A Tableau Alternative in Python for Data Analysis (in Streamlit & Jupyter) | PyGWalker Tutorial

Hey guys, There's a new Python library called PyGWalker,

which lets you turn your Pandas dataframe into a Tableau-style User Interface for visual

analysis.

You can integrate it into your Jupyter Notebook or Streamlit app.

Let me show you how to use it.

First things first, you need to install it.

Just run "pip install pygwalker" in your command prompt or terminal, and you're all set.

Once that's done, you can import it together with pandas.

Next, for demo purposes, let's load the 'tips' dataset, which looks like this.

Here, we can see the total bill amount, along with the tips, gender, whether the customer

was a smoker or not, the day and time of the visit, and the size.

Alright, now let's assume you want to explore this dataset.

Instead of plotting it in different ways, you can simply call "pyg.walk" and pass in

your dataframe.

You could also specify if you want to use a dark or light mode.

After executing this line, you'll get this canvas with two tabs.

Let me first switch to the data tab.

Here, you can inspect your dataframe and check out some entries if you'd like.

You'll also notice that PyGWalker automatically labels the numeric values as measures and

the rest as dimensions.

Now, let's switch back to the visualization tab.

Like in Tableau, you can now drag and drop the available fields into the x and y axes.

So, for example, let me plot the total bill amount by day.

To resize the graph, you can click here and choose the fixed layout mode.

Now you can resize the chart.

By default, PyGWalker picks a chart type for you.

However, you can use other chart types, like a line chart.

But for our analysis, I'll switch back to the bar chart.

The cool thing is, you can drag additional fields into your chart to explore your data

in a very intuitive way.

So, for example, I'll add the tips to the y-axis.

And instead of the total sum, I might want to look at the average bill amount.

If you want to remove a field, you can simply drag it back to the field list.

Likewise, you can add multiple fields to the x-axis.

Let's say, I'm interested to see if smokers spend more money, and I want to see it for

each day.

Additionally, I also want the information if it was during dinner or lunchtime.

To do that, you can use the colour field.

And I'm only interested in this result for male customers.

So you can apply a filter by dragging the column into the filter field.

Now, I'll select only 'male' and confirm my selection.

The values will then be filtered accordingly.

And to remove it, you can either select all again or simply drag it out of the filter

field.

I think it's a very interesting tool for quick exploratory data analysis.

However, you can also export the charts.

So, for example, I'll export it to a scalable vector graphic, and this is what it looks

like.

Alright, now back in my Jupyter Notebook, you can add additional charts by clicking

up here.

And now you have a second canvas, where you can again explore your data.

There are actually a couple more options in PyGWalker, but I would encourage you to play

around with it yourself.

But one important feature is to export your code.

So, if you click the "Export As" button, you'll have different options.

First, let me export it as "code".

Next, I'll hit the 'Copy to Clipboard' button.

Now, if I paste it into a new cell, you'll see the spec of my dashboard.

If I run "pyg.walk" and specify my specs, we'll get back the same dashboard.

Alternatively, you could also export your settings as a file.

If you do so, you'll get a "config.json".

Let me grab this config file from my downloads folder and paste it into my current directory.

Now, if you want, you can also implement the PyGWalker dashboard into your Streamlit app.

To demonstrate this, I've created the following app.

After importing Streamlit, I'll set some basic page configurations.

Next, I'll load our 'tips' dataset and set up a title and subheader.

The interesting part is here.

We can now read the content from the config file and use "pyg.walk" again in our Streamlit

app, together with the previous settings.

All you need to do is specify "Streamlit" in the environment.

This time, I'll also switch it to the dark mode.

Ok, and with that in place, let's spin up my app by running "streamlit run app.py".

Once my page is loaded, we'll have the PyGWalker canvas in our Streamlit app.

And as before, you can now play around with your data directly in Streamlit.

Ok, guys, and that's all I have for you today.

I will upload this Streamlit app and also the Jupyter Notebook to my GitHub repo.

You will find the link in the description box of this video.

As always, thanks for watching, and I'll see you in the next video!