Skip to main content
Version: 0.32.2

Guided Learning

In this section, we'll walk through asking questions with Ava using a dataset from a bikeshare company.

Load Data​

To start, download the dataset "BikeShare dataset (Excel)" from DataChat Training, and load it into a new conversation.

Using the dataset dropdown, we can see that there are three datasets we're asking questions on, "BikeShare", "WeatherDecode", and "SeasonDecode".

loaded data

Investigate Data​

To start, let's investigate our data a bit. Let's say we'd like to know how many records are in this dataset. We can ask "How many records are in the dataset?" and press Enter.

We're then given a result showing us that there are 17,379 records in this dataset.

question 1

We can see how Ava found our solution by clicking the Recipe button.

recipe explanation

Ask More Questions​

From here let's ask a second question: "Which season has the highest average ridership?"

question 2

Ava provides a solution, showing us that fall has the highest average, about 236 riders.

Sweep Conversation​

Let's ask Ava another question. First, let's start a new topic so that our previous conversation history has no bearing on our new results. Click the New Topic button next to the input field, then enter "What most impacts allRiders?".

sweep conversation

Ava automatically begins training a machine learning model to determine which factors most impact allRiders. After some time, we're given a response that includes multiple visualizations:

question 3

Each of the visualizations provided gives us a look into key features that impact allRiders. In this example we can see from the first impact chart that "registeredRiders" and "casualRiders" have the largest impact on all riders, which we expect as those riders are part of the allRiders total. However, we can also see that "hour" also has quite a bit of impact, followed by "temperatureRelativeTo41C and "workingDay".

The charts provided are also interactive. Let's scroll to the "allRiders vs. registeredRidersInt200, splitting on workingDay" violin chart and press Play. The chart cycles through binned groups of casual riders to display allRider counts. We can see that the most casual riders ride during weekends while registered riders tend to ride during working days.

violin chart

Feedback​

Since we're happy with the results that Ava provided, let's give the response positive feedback. Click the Thumbs-Up icon next to the output. This helps Ava to learn and provide better results with each question.

feedback

Share Findings​

We can also create a public link to our responses that we can then share with others. Click the Share icon next to the output. This opens a form where we can select which chart we'd like to share. Let's share the chart we most recently investigated, Violin2, and click Create link. From here you can copy a URL or embed the corresponding code into an external site or dashboard.

share output