Time series X/Y plots

Quiver contains several cards for finding trends by comparing time series values between series or analyzing the frequency of values within the same series.

Quiver supports the following features:

• Creating time series scatter plots by plotting one time series against another.
• Performing linear regression on a time series scatter plot to detect trends.
• Adding numeric formulas based on object properties to scatter plot charts.
• Creating an empirical distribution of time series values over a given time period.
• Creating an empirical distribution of two time series using a heat grid.

Example workflow: Exploring trends in stock data

In this example, we will compare trends in prices between the CMG and AAL stocks. We start by adding both series to our analysis.

Adding a scatter plot

We would like to investigate whether these stock prices are correlated. To test this, we create a scatter plot. For each bucketed timestamp, Quiver generates a point representing the prices of CMG and AAL:

1. Hover over the chart to open the next actions menu.
2. Select Add plot > X/Y plots > Time series scatter plot.
3. Input the AAL time series as the Source X Plot and the CMG time series as the Source Y plot.
4. Creating a scatter plot may be quite expensive for larger time series. Toggle on the Bucketing option and decrease the number of buckets if needed. This will result in fewer points being generated which can improve performance.

Plotting a scatter plot regression

There appears to be a positive correlation between the prices of these two stocks. To quantify this, we can run a scatter plot regression:

1. Hover over the chart to open the next actions menu.
2. Change the card next to Continue analysis from in the menu header from Time Series Chart to the scatter plot from the previous section.
3. Select Visualize > Scatter plot regression.

While we choose to use linear regression for this example, you can also perform exponential and polynomial regression.

We would now like to see whether time plays a role in this trend. For this, we can toggle on the Color points by time option in the Display tab of the scatter plot. Toggling on Show color time legend allows us to see at a glance what time ranges correspond to different colors.

Marking plot averages

Finally, we would like to mark the average prices of CMG and AAL on the chart. To do this, we first use the time series numeric aggregation card to compute the averages. We can use the numeric series formula to draw a horizontal line depicting the average price of CMG:

1. Hover over the chart to open the next actions menu.
2. Select Add plot > X/Y plots > Numeric series formula.
3. Reference the metric card with the average stock price for CMG in the formula.
4. Change the X Unit Label and the Y Unit Label to dollars to match the labels on the scatter plot chart.
5. Drag the numeric series formula plot into the chart with the scatter plot and drag its x-axis into the existing chart x-axis to combine them.

Finally, we add a numeric marker on the chart to draw a vertical line depicting average price of AAL:

1. Select the gear icon in the chart header.
2. Scroll down to the Markers section and select Add marker. Input the metric card with the average stock price for AAL.

Distributions and heat grids

Now, we would like to compare the distributions of the CMG and AAL stocks. We can do this using the time series heat grid and time series distribution cards. To configure the time series heat grid:

1. Hover over the chart to open the next actions menu.
2. Select Add plot > X/Y plots > Time series heat grid.
3. Input the AAL time series as the Source X Plot and the CMG time series as the Source Y plot.

By configuring the number of bins, we can adjust the granularity of the distributions. We can also change the color scheme of the heat grid from the Display tab.

To configure the time series distribution:

1. Hover over the chart to open the next actions menu.
2. Select Add plot > X/Y plots > Time series distribution.
3. Select the AAL time series as the Source plot.
4. Repeat steps 1 - 3 for the CMG time series.
5. To display the two distributions on the same chart, drag one distribution plot into the chart containing the other plot.