paint-brush
No-Code Machine Learning inside Google Sheetsby@navneet-madhu-kumar
3,271 reads
3,271 reads

No-Code Machine Learning inside Google Sheets

by Navneet Madhu KumarMay 22nd, 2020
Read on Terminal Reader
Read this story w/o Javascript
tldt arrow

Too Long; Didn't Read

The Aasaanai Google Sheets Add-On lets you build, train, and use state-of-the-art Machine Learning models, inside Google Sheet, without writing a line of code. We will be adding more tutorials for all those tasks. We also support Forecasting and Text Analysis Models, Forecasting Text Analysis models, Forecast Text Analysis, Text Analysis and Forecast Models. We have all these historic data but leveraging that data for anything more than dashboards is a hard task. Exporting to CSV, Cleaning the data, Normalising, and then tuning your favorite Machine Learning model is a cumbersome process.

Company Mentioned

Mention Thumbnail

Coin Mentioned

Mention Thumbnail
featured image - No-Code Machine Learning inside Google Sheets
Navneet Madhu Kumar HackerNoon profile picture

Introduction

Build, Train, and Use Machine Learning Models inside Google Sheets.

I am a fan of Google Sheets as are, I believe, many others. I use it for almost anything related to my startup, ranging from customer reviews and analytics to website analytics. You could say that Google Sheets is extremely integral to our operations.

We have all these historic data in Google Sheets but leveraging that data for anything more than dashboards is a hard task. Exporting to CSV, Cleaning the data, Normalising, and then tuning your favorite Machine Learning model is a cumbersome process with a steep learning curve.

We, at Aasaanai, found this was a problem shared by many. Many of our fellow startups are looking to leverage the power of AI and Machine Learning but out deterred by the complexity surrounding it.

Hence, we made the Aasaanai Google Sheets Add-On. Now you can build, train, and use state-of-the-art Machine Learning models, inside Google Sheets, without writing a line of code!

Let’s try it now, shall we!

Dataset

We will be using the Car Selling Price Dataset from Kaggle, in this example. Once you download the CSV, just import in Google Sheets.

Get the Add-On

Get the Add-On from the Google Sheets Marketplace.

Setup

There are a few steps that we need to take care of before actually using the Add-On.

For using the Add-On, you need to get an API Key.

Visit the Aasaanai website and press on Get Started.

Now just create an account by pressing the Sign Up button.

Once you have created an account successfully, you will be greeted by this screen.

Move to the upper right corner, to the user logo.

Copy the API Key from here!

Now get back to the Sheets Add-On and select — Set API Key.

Enter the previously copied API key here and the setup is done!

Task: Regression

The task we have here is to predict the selling price of a pre-owned car, given some other parameters.

Go to the Add-on, Press Start, and the Regression.

You will be greeted with the Aasaanai Sidebar. Here you will find all the controls for training and using your models.

Now, click on create your own model.

You will be asked to either add labels and refresh the sheet or directly upload it. Since we already have the Selling Price column, we will press the Upload sheet button.

Once the file upload is done, you will find the training window. Select the Selling Price Column and give the model a name.

Press Train Model. This will start the training process which may take a couple of minutes, depending on the size of your Spreadsheet.

Once the model is trained, you will get a prompt.

Now go back to the homepage, press refresh and you will find the trained model.

Now we can use the model just by clicking on it. You can also delete the model but just clicking the trash button.

Prediction

To use the prediction, we can just click on the model which opens up the prediction window.

Just enter the row starting and ending values. Mind you, sheets indexing starts at 1.

Press predict and voila you will get the results in the sheet.

We also have the option to make triggers. But we will discuss them in another post.

So, Why Aasaan?

Traditional ML Flow

  1. Extract data and convert it to CSV or some other format.
  2. Import into a python environment using Pandas.
  3. Clean dataset by removing unwanted or missing values.
  4. Normalizing the data, deciding on which features are important, and also feature engineering.
  5. Choose the appropriate model and train.
  6. Repeat until you see good performance.

Aasaan

  1. Open Add-On.
  2. Select Label.
  3. Train Model.

Check us out on Aasaanai and the Sheets Add-On. We also support Classification, Forecasting, and Text Analysis Models. We will be adding more tutorials for all those tasks.