Trending December 2023 # Percentage Of Total Using All And Allselected # Suggested January 2024 # Top 18 Popular

You are reading the article Percentage Of Total Using All And Allselected updated in December 2023 on the website We hope that the information we have shared is helpful to you. If you find the content interesting and meaningful, please share it with your friends and continue to follow and support us for the latest updates. Suggested January 2024 Percentage Of Total Using All And Allselected

In this tutorial, I’ll discuss the difference between ALL and ALLSELECTED DAX functions when calculating the percentage of total in Power BI. The difference between these two DAX functions can be relatively confusing when you’re just starting out with Power BI. Hopefully, this tutorial can give you some clarity on this matter. You can watch the full video of this tutorial at the bottom of this blog.

I got this idea from a video that did an introduction about the ALL function. You can check that video from the Enterprise DNA Youtube Channel here. 

In that video, the speaker compared the date versus the total sales using the ALL function. Here, I’m going to take that example one step further and show how to either use the ALL or ALLSELECTED function when calculating the percentage of total sales. This could be by date or by customer.

I’m going to use a Division example in this tutorial.

Basically, Division is like a job type.

I also placed a slicer at the top right part just to show that these results are from year 2023.

And this shows the Invoiced amount for each of the following Divisions.

I also provided a slicer for the Division that we’ll use later once we add the percentage of total invoiced using either the ALL or ALLSELECTED function.

This TREATAS Measures here is where I stored all my invoice measures.

The Invoiced measure is the first measure within my table.

This measure calculates the Invoiced amount, which is the Total Estimates.

I also used the TREATAS function because there’s no relationship between the Date table and the Jobs table, so I created that relationship virtually, instead.

And that’s how I created the Invoiced amount.

Now what I’ll do is to take the Invoiced using the ALL function.

This calculates the sum of all the amount Invoiced using the Invoiced measure that I previously discussed. I also used the ALL function to display all the results by Division in the Jobs table.

By adding the Invoiced ALL measure to this table, it only displays the total amount of invoice for each one of these rows.

So, that’s what the ALL function does. It returns all the rows in a table, or all the values of a column while ignoring any existing filter that might have been applied.  

After adding the Invoiced ALL measure to the table, the next thing that I want to do is to show the percentage of total sales for each one of these Divisions for the year of 2023. 

To do that, I created another measure which I named as ALL Invoiced%. In this measure, I just divided the Invoiced measure by the Invoiced ALL measure.

Then, I’ll add that measure to the table. As you can see, it’s actually working correctly based on the results for Reconstruction Division. It shows that it has $775,766 out of $1,866,767, which makes sense for a percentage total of 41.56%.

But what if I only want to select a certain Division?

For example, I’ll use my slicer here so the table will only display the Reconstruction and the Mold Remediation divisions.

Noticeably, the ALL Invoiced% column is still displaying the same percentage.

It’s not showing the expected results that I want. This is because it’s basically just taking the Invoiced result divided by the Invoiced ALL result to get the percentage value.

What I want is to show the percentage of the Reconstruction and Mold Remediation out of the current total Invoiced amount.

This is where the ALLSELECTED function comes in.

I’ll unselect the Reconstruction and Mold Remediation selections for now. Then, let’s check out another measure that I created for Invoiced using the ALLSELECTED function. I named it Invoiced ALLSELECTED.

In this measure, I used the measure branching technique again. But instead of using the ALL function, I used the ALLSELECTED function.

I’ll add that measure again to the table. As you can see, the Invoiced ALLSELECTED column is showing the same amount as Invoiced ALL.

This is because by default, all the Divisions are selected in this model and I haven’t used the slicer yet.

I also created a measure named ALLSELECTED Invoiced% to get the percentage of total sales for each one of these Divisions for the year of 2023.

It’s similar to the ALL Invoiced% measure, but I used the ALLSELECTED function here instead of the ALL function.

Upon adding that to the table, you’ll see that it’s showing similar results from the ALL Invoiced% column.

However, here’s where the trick of this tutorial comes in. I’ll use the Division slicer again and select Reconstruction and Mold Remediation.

And you’ll see that the result of the ALLSELECTED Invoiced% column is now different from the ALL Invoiced% column.

The ALL Invoiced% column is only displaying 44.40%, because it’s still calculating the Invoiced amount of the other divisions even though they’re not selected.

On the other hand, the ALLSELECTED Invoiced% column where we used the ALLSELECTED function displays a 100% total. This is because it’s only calculating the Invoiced amount of the selected divisions.

This correctly shows that the Mold Remediation division makes 6% and the Reconstruction division makes up the 93% and a half of the $828,925 current total of Invoiced from both divisions.

To sum up, this is the difference between the ALL and the ALLSELECTED function. In this example, I’ll select more Division to further see the difference.

After selecting the Water Mitigation division, the numbers under the ALLSELECTED Invoiced% and ALL Invoiced% columns displayed a noticeable change.

That’s all I wanted to share in this tutorial. This valuable tip can definitely help you in calculating the correct percentage of total, whether it may be invoiced or total sales. Moreover, I hope this tutorial has given you the clarity on the difference between the ALL and ALLSELECTED functions in Power BI.

Check out the links below and our website as well for more examples and related content.


You're reading Percentage Of Total Using All And Allselected

Return On Total Assets Formula

Return on Total Assets Formula (Table of Contents)

Start Your Free Investment Banking Course

Download Corporate Valuation, Investment Banking, Accounting, CFA Calculator & others

What is Return on Total Assets Formula?

The term “Return on Total Assets” refers to the financial ratio used as an indicator to check how well a company can use its assets to generate earnings during a specific period. In other words, it measures the profitability of the company’s available assets. The Return on Total Assets can be derived by dividing the company’s earnings before interest and taxes (EBIT) by its average total assets. Mathematically, it is represented as,

Return on Total Assets = EBIT / Average Total Assets

Examples of Return on Total Assets Formula (With Excel Template)

Let’s take an example to understand the calculation of the Return on Total Assets in a better manner.

You can download this Return on Total Assets Formula Excel Template here – Return on Total Assets Formula Excel Template

Return on Total Assets Formula – Example #1

Let us take the example of a company with reported earnings before interest and taxes (EBIT) of $75,000 as per the income statement. As per the balance sheet for the year ending on December 31, 2023, the average total assets of the company stood at $5,000,000. Calculate the Return on Total Assets for the company during the period.


The formula to calculate Return on Total Assets is as below:

Return on Total Assets = EBIT / Average Total Assets

Return on Total Assets = $75,000 / $5,000,000

Return on Total Assets = 1.50%

Therefore, the company reported a Return on Total Assets of 1.50% during the period.

Return on Total Assets Formula – Example #2

Let us take the example of ABC Ltd, which reported a net profit of $50,000 on a turnover of $500,000. Per its income statement, the interest expense and income taxes stood at $15,000 and $30,000, respectively. Further, as per the balance sheet, the opening and closing value of the total assets is $3,900,000 and $4,100,000, respectively. Calculate the Return on Total Assets for ABC Ltd based on the latest reported financials.


The formula to calculate EBIT is as below:

EBIT = Net Income + Interest Expense + Income Taxes

EBIT = $50,000 + $15,000 + $30,000

EBIT = $95,000

The formula to calculate Average Total Assets is as below:

Average Total Assets = (Opening Total Assets + Closing Total Assets) / 2

Average Total Assets = ($3,900,000 + $4,100,000) / 2

Average Total Assets = $4,000,000

The formula to calculate Return on Total Assets is as below:

Return on Total Assets = EBIT / Average Total Assets

Return on Total Assets = $95,000 / $4,000,000

Return on Total Assets = 2.375%

Therefore, ABC Ltd managed a Return on Total Assets of 2.375% during the last reported year.

Return on Total Assets Formula – Example #3

Take the real-life example of Apple Inc., which reported a net income of $59,531 Mn during the last reported financial year. As per the annual report for the year ending September 29, 2023, the interest expense and provision for income taxes for the year stood at $3,240 Mn and $13,372 Mn, respectively. Further, the total asset at the beginning and end of the year stood at $375,319 Mn and $365,725 Mn, respectively. Calculate the Return on Total Assets for Apple Inc. based on the information.


The formula to calculate EBIT is as below:

EBIT = Net Income + Interest Expense + Income Taxes

EBIT = $59,531 Mn + $3,240 Mn + $13,372 Mn

EBIT = $76,143 Mn

The formula to calculate Average Total Assets is as below:

Average Total Assets = (Opening Total Assets + Closing Total Assets) / 2

Average Total Assets = ($375,319 Mn + $365,725 Mn) / 2

Average Total Assets = $370,522 Mn

The formula to calculate Return on Total Assets is as below:

Return on Total Assets = EBIT / Average Total Assets

Return on Total Assets = $76,143 Mn / $370,522 Mn

Return on Total Assets = 20.55%

Therefore, Return on Total Assets for Apple Inc. stood at 20.55% for the year ending on September 29, 2023.


The formula for Return on Total Assets can be derived by using the following steps:

Step 1: First, calculate the company’s net income from its income statement. Next, determine the interest expense incurred and corporate taxes paid during the year. Add the interest expense and tax to the net income to compute the company’s EBIT.

EBIT = Net Income + Interest Expense + Tax

Step 2: Next, determine the company’s total assets at the beginning and the end of the current year. The total assets include short-term and long-term assets for the period under consideration. Now, add the values for total assets and divide by 2 to arrive at the average total assets.

Average Total Assets = (Opening Total Assets + Closing Total Assets) / 2

Step 3: Finally, the formula for Return on Total Assets can be derived by dividing the company’s EBIT (step 1) by its average total assets (step 2), as shown below.

Return on Total Assets = EBIT / Average Total Assets

Relevance and Uses of Return on Total Assets Formula

It is one of the important profitability metrics that allows an analyst to assess the effectiveness of a company in its asset utilization. A higher Return on Total Assets value indicates favorable healthy asset utilization to produce greater earnings, eventually attracting investors. Inherently, a positive ratio signifies an upward trend for profit.

The ratio can be used to compare companies of the same scale and in a similar industry. However, comparing companies from different industries is meaningless as asset utilization varies significantly.

Return on Total Assets Formula Calculator

You can use the following Return on Total Assets Calculator

EBIT Average Total Assets Return on Total Assets Formula   Return on Total Assets Formula = EBIT =

Average Total Assets


= 0


Recommended Articles

This is a guide to the Return on Total Assets Formula. Here we discuss calculating the Return on Total Assets and practical examples. We also provide a Return on Total Assets calculator with a downloadable Excel template. You may also look at the following articles to learn more –

All Of The Zoom Keyboard Shortcuts And How To Use Them

With more people working from home full-time, it’s common to have Zoom open all day. But you might not realize that there are dozens of shortcuts that can improve your user experience and boost your efficiency.

In this article, we’ll cover all of the Zoom shortcuts for Windows, Mac, Linux, and iOS, as well as how to use them.

Table of Contents

Zoom Shortcuts for Windows, Mac, and Linux

Zoom has various shortcuts available for every supported platform. These accessibility settings are designed to save time and effort in Zoom meetings.

For shortcuts to work on Windows, you must be using the Zoom desktop client version 5.2.0 or higher. Additionally, all keyboard shortcuts can be viewed and customized. To change your shortcuts:

Select any shortcut and press the key you would like to use for it.

With that out of the way, here are the default shortcuts:

General Shortcuts

To switch between open Zoom windows, press F6 on Microsoft Windows, Ctrl + T on Mac, and Ctrl + Tab on Linux.

To shift focus to Zoom’s meeting controls, press Ctrl + Alt + Shift on Windows.

Meeting Shortcuts

Hold key to talk while muted: Spacebar on Windows, Linux, and Mac.

Show or hide meeting controls: Alt on Windows and Linux, and Ctrl + / on Mac (this toggles the Always show meeting controls option).

Switch to the active speaker view: Alt + F1 on Windows and Command + Shift + W on Mac (depending on the current view).

Switch to the gallery view: Alt + F2 on Windows and Command + Shift + W on Mac (depending on the current view).

Close the current window: Alt + F4 on Windows and Command + W on Mac.

Start/stop video: Alt + V on Windows and Linux, and Command + Shift + V on Mac.

Unmute or mute audio: Alt + A on Windows and Linux, and Command + Shift + A on Mac.

Mute or unmute audio for everyone except for the host (only available to the meeting host): Alt + M on Windows and Linux, and Command + Control + M on Mac (and Command + Control + U to unmute).

Share screen (meeting controls need to be in focus): Alt + S on Windows and Linux, and Command + Control + S on Mac.

Pause or resume screen sharing (meeting controls need to be in focus): Alt + T on Windows and Linux, and Command + Shift + T on Mac.

Start or stop local recording of the meeting: Alt + R on Windows and Linux, and Command + Shift + R on Mac.

Start or stop cloud recording: Alt + C on Windows and Linux, and Command + Shift + C on Mac.

Pause or resume recording: Alt + P on Windows and Linux, and Command + Shift + P for Mac.

Switch camera: Alt + N on Windows and Linux, and Command + Shift + N on Mac.

Toggle fullscreen mode: Alt + F on Windows, Command + Shift + F on Mac, and Esc on Linux.

Toggle the in-meeting chat panel: Alt + H on Windows and Command + Shift + H on Mac.

Show or hide participants panel: Alt + U on Windows and Linux, and Command + U on Mac.

Open invite window: Alt + I on Windows and Linux, and Command + I  on macOS.

Raise or lower hand in the meeting: Alt + Y on Windows and Linux, and Option + Y on Mac.

Read the active speaker’s name: Ctrl + 2 on Windows.

Toggle floating meeting control toolbar: Ctrl + Alt + Shift + H on Windows and Ctrl + Option + Command + H on Mac.

End or leave meeting: Alt + Q on Windows and Command + W on Mac.

Gain remote control: Alt + Shift + R on Windows and Linux, and Control + Shift + R on Mac.

Stop remote control: Alt + Shift + G on Windows and Linux, and Control + Shift + G on Mac.

View the previous 25 video streams in gallery view: PageUp in Windows.

View the next 25 streams in gallery view: PageDown in Windows.

Chat Shortcuts

Take a screenshot: Alt + Shift + T on Windows and Linux, and Command + T on Mac.

Toggle portrait or landscape view: Alt + L on Windows and Command + L on Mac.

Close current chat: Ctrl + W on Windows and Linux.

Open previous chat: Ctrl + Up on Windows.

Open the next chat: Ctrl + Down on Windows.

Jump to the chat window: Ctrl + T on Windows and Command + K on Mac.

Search within the chat: Ctrl + F on Windows.

Phone Call Shortcuts

Accept the inbound call: Ctrl + Shift + A on Windows, Linux, and macOS.

End the current call: Ctrl + Shift + E on Windows, Linux, and macOS.

Decline the inbound call: Ctrl + Shift + D on Windows, Linux, and macOS.

Mute or unmute microphone: Ctrl + Shift + M on Windows, Linux, and macOS.

Hold or unhold current call: Ctrl + Shift + H on Windows, Linux, and macOS.

Call the number highlighted: Ctrl + Shift + P on Windows and Ctrl + Shift + C on Mac.

Zoom Shortcuts for iOS

The iOS Zoom app also has a handful of shortcuts that you can use if you’re accessing Zoom from an iPad or iPhone with a keyboard. These are:

Command + Shift + A: Mute or unmute audio.

Command + Shift + V: Start or stop video.

Command + Shift + H: Display or hide chat.

Command + Shift + M: Minimize the meeting.

Command + U: Toggle participants list.

Command + W: Close the participants or settings window (whichever is open).

Taking Efficiency to the Next Level

That’s every Zoom keyboard shortcut for the Windows, Mac, Linux, and iOS apps. With these hotkeys, you can improve your overall user experience, save time, and become a videoconferencing pro.

Evolution Of Malware – How It All Began!

Hi. I am Creeper. Catch me if you can. It was the ’50s! Back then, computers were big. Programmers used punching cards. One such programmer – Bob Thomas – experimented with self-replicating programs and created the Creeper. Fortunately, the worm could not self-replicate, but it did affect users of ARPAnet (one of the first computer networks’ community). From there, began a journey into the most dangerous realms of the Internet.

Let us check out the evolution of malware after taking a quick look at the difference between a worm and a virus.

Worm vs Virus

A Worm is basically a program that can self-replicate across computers and other types of digital devices. A Virus needs to be attached to something like an application and needs a trigger, such as the execution of that application, to work for whatever intention it was created. In other words, worms are independent and can replicate without the need for any triggers. They can be downloaded with other programs. They can affect your computers bypassing through Flash drives. A virus uses plenty more methods in addition to the two mentioned here, to get into a computer and infect it. It requires some action to be taken by the user before it becomes active and does the work for which it is programmed.

These days, we do not hear about worms explicitly. We have a common word called Virus and even a more generic one – Malware. Since these days, the intentions of worms and viruses, plus other types of software like Spyware, etc. are malicious or bad, they are collectively called Malware. Unlike the beginning, where malware was the result of curiosity and experimentation, and the intention was merely to irritate, cause mischief or havoc, these days’ viruses are full programs that are intended to steal or destroy data. The intentions are bad as the industry creates malware for their benefits at your costs.

Evolution of Malware and Viruses

Although WinVer 1.4 was said to be the first Windows virus, the first malware to be introduced to the world was the Creeper. It was not a malware by definition though. It simply displayed a message which would irritate users and as a result, the first anti-virus software was born. It was named Reaper and it was made to counter Creeper. There are different arguments saying this cannot be called malware as it could not replicate or cause damage to computers, but still, many accept Bob Thomas and his Creeper as the beginning of what later turned into a multi-billion industry of malware. Bob could not have even imagined that.

Anyway, the next malware was said to be Brain. It was developed by two Pakistan-based people in 1986. By this time, the general public too had fans of computers and there were many hobby groups and communities that were run using computers. The target of Brain was these communities. It targeted the boot sector of computers via a 5 1/4 inch floppy disks and showed just a message. It, too, was not intended to steal data or cause data loss in any way. It also gave the phone number of the malware developers – Basit and Amjad – so that people could ask them for help to remove the malware.

The first reference to a worm that caused damage (presumed to be because of a bug in the worm code) was Morris’ worm. It was developed by Robert Morris, a student at Carnell University. Again, as with the Creeper, people argued this was the first worm – as it could replicate. “Worms need to replicate else they are not worms”, people argue. This infected more than 5000 computers in the USA and caused damage between 100,000 and 10,000,000. The exact damage could not be estimated.

The biggest turn in the history of malware or its evolution was the LoveLetter worm. By that time, most organizations had computers working on MS-DOS or other similar operating systems. It was the year 2000 and the LoveLetters that contained an infected attachment which when, downloaded, infected the email program and sent a copy of the worm to people in recipients’ address book. Not only that, it overwrote certain file types with rubbish. By the time it was discovered as not being a prank and a serious threat, the damage was done. However, it educated people about malware and that people out there are not all good – but bad ones too who would want to play with the data they had on their computers.

A need for anti-virus software on every computer was stressed and was implemented slowly. Of course, those were small codes that kept on updating themselves as and when new worms or viruses are discovered.

The year 2001 saw the emergence of Red Code, a malware that targeted Microsoft IIS based systems. Normal antivirus could not find it as it was resident in the active memory of the computer. The worm could be detected only in transit. Traditional antivirus failed and the need arose for better ones that can scan all parts of a computer where such malware can reside: boot sector, memory, hard disks, application files, etc.

Then came Win32/Ninda which was a threat to Networks. It used network backdoors to spread and affected hundreds of thousands of computers and web servers. Many websites were compromised and provided as a source for further infections. By this time, Internet usage was in full swing. It is said the malware initiated around the attacks of Sept 11, 2001. Antivirus vendors went back to their drawing boards to create antivirus that could also monitor network ports, especially Port 80 – the one used to connect to the Internet and detection of other open or closed ports that they need(ed) to hide from the networks.

People were also educated about the possibilities of Spyware, Adware, etc and the collective term, Malware, was subsequently coined. You can read the difference between Virus, Trojan, Worm, Adware, Rootkit, etc, here.

Over the last two decades, both malware and anti-malware programs have become complex. Phishing became part of the Internet soon and antivirus had to scan complete emails – including the contents – to make sure there are no malicious URLs, etc.

We can say that in the last decade, especially, had seen a tremendous rise in dreaded virus problems, as well as good improvements in the antimalware solutions. There are many free antivirus software and free Internet Security Suites, that act as well as the paid options. One now needed to take an integrated approach to fight malware, and hence Firewalls, Heuristics, etc, were also made a part of the arsenal.

There are competing claims for the innovator of the first antivirus product. Possibly the first publicly documented removal of a computer virus in the wild was performed by Bernd Fix in 1987. By the end of 1990, there were a number of anti-virus products available.


If you’d like to find out more about how malware grew in time, download this PDF copy of the Malware History whitepaper from BitDefender. There is also a lot of information at Microsoft, on the evolution of malware and malware trends.

Ransomware, Rogue software, Rootkits, Botnets, RATs, Malvertising, Phishing, Drive-by-download attacks, Online Identity Theft, are all here to stay now. New technologies that have emerged or are emerging, including but not limited to BYOD and the Internet of Things will be attacked. Malware has also started focusing on Social Media. While good security software will help you stay protected, it is equally important to carry out safe Internet and Browsing practices.

Classification Of Handwritten Digits Using Cnn

This article was published as a part of the Data Science Blogathon


(CNN) for classifying handwritten digits from a popular dataset.

Figure 1: MNIST Dataset (Picture credits: 


Although each step will be thoroughly explained in this tutorial, it will certainly benefit someone who already has some theoretical knowledge of the working of CNN. Also, some knowledge of TensorFlow is also good to have, but not necessary.

Convolutional Neural Network

For those of you new to this concept, CNN is a deep learning technique to classify the input automatically (well, after you provide the right data). Over the years, CNN has found a good grip over classifying images for computer visions and now it is being used in healthcare domains too. This indicates that CNN is a reliable deep learning algorithm for an automated end-to-end prediction. CNN essentially extracts ‘useful’ features from the given input automatically making it super easy for us!

Figure 2: End to end process of CNN

A CNN model consists of three primary layers: Convolutional Layer, Pooling layer(s), and fully connected layer.

(1) Convolutional Layer: This layer extracts high-level input features from input data and passes those features to the next layer in the form of feature maps.

(2) Pooling Layer: It is used to reduce the dimensions of data by applying pooling on the feature map to generate new feature maps with reduced dimensions. PL takes either maximum or average in the old feature map within a given stride.

(3) Fully-Connected Layer: Finally, the task of classification is done by the FC layer. Probability scores are calculated for each class label by a popular activation function called the softmax function.

For more details, I highly recommend you check this awesome tutorial on Analytics Vidhya.


The dataset that is being used here is the MNIST digits classification dataset. Keras is a deep learning API written in Python and MNIST is a dataset provided by this API. This dataset consists of 60,000 training images and 10,000 testing images. It is a decent dataset for individuals who need to have a go at pattern recognition as we will perform in just a minute!

When the Keras API is called, there are four values returned namely- x_train, y_train, x_test, and y_test. Do not worry, I will walk you through this.

Loading the Dataset

The language used here is python. I am going to use google colab for writing and executing the python code. You may choose a jupyter notebook as well. I choose google colab because it provides easy access to notebooks anytime and anywhere. It is also possible to connect a colab notebook to a GitHub repository.

Also, the code used in this tutorial is available on this Github repository. So if you find yourself stuck someplace, do check that repository. To keep this tutorial relevant for all, we will understand the most critical code.

Create and name a notebook


After loading the necessary libraries, load the MNIST dataset as shown below:

(X_train, y_train) , (X_test, y_test) = keras.datasets.mnist.load_data()

As we discussed previously, this dataset returns four values and in the same order as mentioned above. Also, x_train, y_train, x_test, and y_test are representations for training and test datasets. To get how a dataset is divided into training and test, check out the picture below which I used during a session where I talked about C

Figure 3: Dividing the dataset into training and test set

Voilà! You just loaded your dataset and are ready to move to the next step which is to process the data

Processing the Dataset

Data has to be processed, cleaned, rectified in order to improve its quality. CNN will learn best from a dataset that does not contain any null values, has all numeric data, and is scaled. So, here we will perform some steps to ensure that our dataset is perfectly suitable for a CNN model to learn from. From here onwards till we create CNN model, we will work only on the training dataset. 

If you write X_train[0] then you get the 0th image with values between 0-255 (0 means black and 255 means white). The output is a 2-dimensional matrix (Of course, we will not know what handwritten digit X_train[0] represents. To know this write y_train[0] and you will get 5 as output. This means that the 0th image of this training dataset represents the number 5. 

So, let’s scale this training and test datasets as shown below:

X_train = X_train / 255 X_test = X_test / 255

After scaling, we should convert the 2-d matrix to a 1-d array by using this:

X_train = X_train.reshape(-1,28,28,1)  #training set X_test = X_test.reshape(-1,28,28,1)      #test set

Now that the dataset is looking good, it is high time that we create a Convolutional Neural Network.

Creating and Training a CNN

Let’s create a CNN model using the TensorFlow library. The model is created as follows:

convolutional_neural_network = models.Sequential([ layers.Conv2D(filters=25, kernel_size=(3, 3), activation='relu', input_shape=(28,28,1)), layers.MaxPooling2D((2, 2)), layers.Conv2D(filters=64, kernel_size=(3, 3), activation='relu'), layers.MaxPooling2D((2, 2)), layers.Conv2D(filters=64, kernel_size=(3, 3), activation='relu'), layers.MaxPooling2D((2, 2)), layers.Flatten(), layers.Dense(64, activation='relu'), layers.Dense(10, activation='softmax') ])

Take some time to let this entire code sink in. It is important that you understand every bit of it. In the CNN model created above, there is an input layer followed by two hidden layers and finally an output layer. In the most simpler terms, activation functions are responsible for making decisions of whether or not to move forward. In a deep neural network like CNN, there are many neurons, and based on activation functions, neurons fire up and the network moves forward. If you do not understand much about activation functions use ‘relu’ as it is used most popularly.

Once the model has been created, it is time to compile it and fit the model. During the process of fitting, the model will go through the dataset and understand the relations. It will learn throughout the process as many times as has been defined. In our example, we have defined 10 epochs. During the process, the CNN model will learn and also make mistakes. For every mistake (i.e., wrong predictions) the model makes, there is a penalty and that is represented in the loss value for each epoch (see GIF below). In short, the model should generate as little loss and as high accuracy as possible at the end of the last epoch.

GIF 1: Training CNN and the improved accuracies during each epoch

Making Predictions

To evaluate the CNN model so created you can run:

convolutional_neural_network.evaluate(X_test, y_test)

It is time to use our test dataset to see how well the CNN model will perform.

y_predicted_by_model = convolutional_neural_network.predict(X_test)

The above code will use the convolutional_neural_network model to make predictions for the test dataset and store it in the y_predicted_by_model dataframe. For each of the 10 possible digits, a probability score will be calculated. The class with the highest probability score is the prediction made by the model. For example, if you want to see what is the digit in the first row of the test set:


The output will be something like this:

array([3.4887790e-09, 3.4696127e-06, 7.7428967e-07, 2.9782784e-08, 6.3373392e-08, 6.1983449e-08, 7.4500317e-10, 9.9999511e-01, 4.2418694e-08, 3.8616824e-07], dtype=float32)

Since it is really difficult to identify the output class label with the highest probability score, let’s write another code:


And with this, you will get one of the ten digits as output (0 to 9).


In this blog, we begin by discussing the Convolutional Neural Network and its importance. The tutorial also covered how a dataset is divided into training and test dataset. As an example, a popular dataset called MNIST was taken to make predictions of handwritten digits from 0 to 9. The dataset was cleaned, scaled, and shaped. Using TensorFlow, a CNN model was created and was eventually trained on the training dataset. Finally, predictions were made using the trained model.


The media shown in this article are not owned by Analytics Vidhya and are used at the Author’s discretion.


G Data Total Security Review: The Best Antivirus App You’ve Never Heard Of

G Data is a well-organized and capable antivirus suite from Germany. It offers two malware engines for added protection, and is priced well. Although it’s not as well known as other antivirus suites, G Data does a good job and the Total Security offering has plenty of features for those who want more than malware scans out of their security suite.

Nothing says security and trust like German quality. At least that’s the pitch for Germany-based G Data. The company’s tag line is “Trust in German Sicherheit (safety).” The idea being that modern Germany is known for “solid German quality” and strict privacy laws, which G Data must adhere to, thereby shielding its customers from malware and privacy-busting breaches.

Note: This review is part of our 

best antivirus

roundup . Go there for details about competing products and how we tested them.


G Data Total Security uses two malware engines.

Similar to other legacy security companies like McAfee and Symantec, G Data maintains a fairly well-organized suite. Its top product, G Data Total Security, has enough to please users who desire a feature-packed suite, while still being simple enough that you’re not overwhelmed. 

G Data Total Security is priced at $50 for a single device for one year, or $82 per year for five devices, and $122 for 10 devices. G Data’s pricing isn’t bad, but it’s just a tad more expensive than other mainstream suites at the 10-device level. G Data’s pricing only covers PCs and Mac, whereas many other suites throw in mobile coverage as part of the plan. G Data’s mobile app for Android is sold separately at $16 per year for a single device, and there’s a free version as well. 


When you first start G Data Total Security for Windows it displays a dashboard called the SecurityCenter with your system’s current protection status. Like many other security suites, it uses a color-coding system; if everything is green in the SecurityCenter you’re good to go.


G Data Total Security’s Virus protection section.

The Virus protection section is where you can manage your virus scanning schedule or start a manual scan. There’s also an option to check for deep-level malware by scanning system folders, RAM, startup files, and doing a rootkit check. You can also view quarantined files here, and create a bootable drive to scan your computer for viruses. The latter option is a great idea as a backup measure—it will save the day should you ever get hit with a particularly nasty bit of malicious software.


Total Security’s Tuner is highly customizable.

Tuner contains the usual antivirus “extras” that help you optimize your system by clearing out temporary files, and so on. The Tuner also bumps up security by disabling potential vulnerabilities like script execution and JavaScript in Adobe Reader. The nice thing about Tuner is that all the actions it takes are listed in checkbox format, allowing you to turn off the things you don’t want to run. 

The Encryption option lets you put sensitive documents in an encrypted container. The Autostart manager tab is just a slightly easier interface for controlling which programs begin at startup. Windows 10 users, however, don’t really need this as the Task Manager can accomplish the same thing.


Total Security’s Device control lets you restrict who can save files to connected drives.

Finally, Device control lets you regulate how users on the PC can access connected drives. Mom and Dad could be allowed to store files on an entertainment content drive, for example, while the kids would have read-only access.

Diving into settings, there isn’t a whole lot you need to adjust. By default, G Data offers to scan flash drives inserted in your USB ports. There’s also a USB Keyboard Guard that protects against USB devices that may pose as a keyboard and try to deliver malware to your PC surreptitiously.


G Data performed quite well in AV-Test’s evaluations. In November and December 2023, G Data’s lower-tier Internet Security scored 100 percent against 216 samples of zero-day, and web and email threats. The larger test with more than 11,000 samples of widespread and prevalent malware also scored 100 percent.

Over at AV-Comparatives, G Data blocked 99.6 percent of threats in the real-world protection test for July through October 2023, with 10 false positives. That score put it just barely behind F-Secure, Panda, Total Defense, Total AV, Trend Micro, and Vipre, but in the same league as pretty much every other major suite including Avast, AVG, McAfee, Norton, and others.

In the malware protection test for September 2023 at AV-Comparatives, G Data nailed it with 100 percent, blocking more than 10,000 samples, with six false positives. 


G Data Total Security gives you the option to scan USB drives.


G Data is a fine antivirus suite. It’s really easy to use, comes loaded with features, and is priced well.

Editor’s note: Because online services are often iterative, gaining new features and performance improvements over time, this review is subject to change in order to accurately reflect the current state of the service. Any changes to text or our final review verdict will be noted at the top of this article.

Update the detailed information about Percentage Of Total Using All And Allselected on the website. We hope the article's content will meet your needs, and we will regularly update the information to provide you with the fastest and most accurate information. Have a great day!