Since we’ve simplified the ML process, one of the main challenges businesses face is data collection and taking action from data.
As data collection tools are becoming easier to come by and businesses are understanding the value of data, few are understanding what to do with their data.
One of the easiest and ~fun~ ways to collect data is this nifty tool called Typeform. You may have heard of it. They hail from Barcelona, Spain and have totally disrupted the form space. Now you can collect data from your audience in an interactive way. And because they have tons of integrations, it’s incredibly easy to store Typeform responses in a database or Google Sheet.
Let us show you how! If you haven’t created a Typeform yet, we recommend you do so this post makes a whole lot more sense.
How to Format Your Dataset
Hopefully you have collected some responses.
Inside your Typeform account, you can quickly turn your responses into a CSV (comma separated values) file. This type of file is very common in data science and is what you’ll be using to make predictions.
In the “Connect” tab under “Integrations” you’ll see a button to connect you to your Google Sheets account. Go ahead and click it to get started. It should prompt you to input already collected responses into a spreadsheet.
Once your data is inside the sheet, you just need to do some light cleaning.
Here are some sample results we received from a Typeform survey with made-up customer IDs. It’s important to note that the more data you have the better. You want to have over 1,000 rows and 5 columns for a decent prediction. Because ML identifies relationships between your columns, you want as many as you can. Of course, this is sometimes tricky if you want a short survey since a column equals a question. However, if you can get your audience to engage and complete your Typeform, the better prediction you will have.
Here are some best practices for collecting Typeform data to use for predictions:
Have an Identifier Column
Have an identifier column so you can make predictions on each respondent. Typeform allows you so ask for an email or name. This can be used for an identifier column. When you transform your responses into a dataset, you can change the names to numbers if you’re going to share this dataset or the prediction outcomes to protect personal information.
Make Your Column Names Simpler
When you transform your data into a Google sheet, each question in your Typeform will come out as its own column. Before uploading your spreadsheet into Obviously AI, we recommend shortening the questions to 1-3 words for simplicity.
Here is an example taken from a public mental health in tech survey. Creating a list like this is important to understanding the dataset, sharing it with your team, and makes understanding the outcomes easier.
Avoid Open Response Questions
When creating a Typeform, you can create open response questions where the participant can type long answers. While this is great for getting qualitative data from your audience, this type of response isn’t great for data predictions. We recommend you stick to uploading quantitative data instead.
When you receive your prediction report, Obviously AI puts the data into an X and Y axis to help you better understand the relationship between two variables. Using a paragraph of text won’t work in this instance because you can’t identify relationships between quantitative and large bodies of text.
If you did collect open answer responses, that’s okay! Tuck them away into your hard drive, but exclude them from the dataset you’ll use to make predictions.
Turning Typeform Data Into a CSV File
Once your dataset is in a good place, you can download the dataset into a CSV (comma separated value) file. You’ll use this file to upload into Obviously AI.
Reminder: If you’ve created a huge amount of data, you can save your CSV to a Dropbox folder and upload files larger than 25MB into Obviously.
To do this, click the “File” button in the top left corner, “Download,” then “CSV.”
Uploading Data to Obviously AI and Making Predictions
Now you’re ready to make data predictions!
Login to Obviously AI and hit “Add Dataset.” Your Data Dialog will appear. Here we predicted satisfaction. Remember when we discussed the Identifier Column. This is where you need one. The customer name or ID number should be used in the Identifier column. Feel free to add any filters or predict a certain attribute here.
Don’t Just Collect Data, Take Action From It
You should be all ready to go with making data predictions from Typeform data.
You’ve learned how to:
- Turn your Typeform responses into a spreadsheet
- Format your data
- Turn your spreadsheet into a CSV file
- Make predictions from your Typeform data
To make it easier, we have a Typeform integration in the works! 😉 In the meantime check out our other integrations here.