Sign in to continue learning
Python for Data Analytics
45 hours 3 mins
You learned how to conduct a ton of cross-sectional analyses in Python in the previous modules. So what are your options for time series data? You can start with visualization. To recap, you've imported your modules and read in your data set into the variable df. Then you created a subset of your data to focus on, apple stocks. You set your indices to date times to create a time series. And finally, you plotted the adjusted close in a line chart. Adjusted close price for apple stocks in February 2020. Now that you have a line chart, you can adjust its appearance. Add a y-axis label and a title. A larger font size will help with legibility. It's pretty clear these values on the x-axis are dates, so you can leave them without an axis title. You'll see how to format these axes more in the next video. You can adjust the appearance of this line as well, using arguments to sns.lineplot, like line style, color, line width, and so on. Since you want to focus on February, when news of COVID broke, you can use the color red and use the dashed line style. Now that you've created this nice looking graph, you may want to plot multiple series on the same chart. Right now you're focusing on February 2020, but you can add some more context to this analysis by including the series for January and March too. You can add all these lines to the same chart. Start with January. Select the January stocks using the dates for January. Remember, these bounds are inclusive, so even though January 1st is a holiday for the U.S. stock market, you can include that value in your slice, and you can go all the way up to January 31st. Then, as you saw previously, you can stack multiple axes on top of each other in the same chart. So copy your code from the previous chart. Change the title, Jan through February 2020, then just add another sns.lineplot call, this time with the January 2020 data. What's wrong with this plot? Well, it's behaving as expected. Since you're plotting a series, you have the date on the x-axis, and these dates don't match up. But wouldn't it be nicer to compare the behavior of each month more directly? That would mean having the same values on the x-axis. So, these adjusted closed prices are the y-values. Which would you set as the x-values? You can use the day from the index. Remember, the index is a time series, so you can take the filtered data frame, dot index, dot day. If you look at that value independently, by the way, it's a series of numbers. There's some missing, but don't worry, Seaborn will handle those gracefully. So, update the x-values for both series. Maybe January should be a solid black line. Cool. Now you can add your series for March. Select your data. Maybe dark gray will work best for the line. Again, you want to highlight February in your analysis, so these other lines can be plotted in more muted colors. Finally, update the title in your graph. So, you can see that February saw a sharp decline, but it was nothing compared with March 2020. Perhaps you should adjust the coloring in your graph to reflect that. Now, finally, just by looking at this graph, your clients won't be able to tell which line is which. Now, finally, just by looking at this graph, your clients won't be able to tell which line is which. If you try using plot legend, you get this tiny empty box, which is not what you need. Why not ask your LLM how to add labels in a legend for each month? Okay, copy that code. Remember, for visualizations in particular, if the code works, it works. Cool. The legend is positioned well. The only thing you may want to do is remove the year, which is in the title already. Recapping what you just learned, you saw that you can use the sms.lineplot function to create a line chart of time series data. You can plot a series of data directly, or you can use the x and y named arguments for more control over the data. You saw that you can use the index, if it's a date time, to plot different values on the x-axis, like the day of the month using .index.day. You also saw that you can use the label named argument to add a label to each line before calling plot.legend. Line charts are your go-to visualization for time series data. However, formatting dates can be a little tricky. Follow me to the next video to see how to approach this task.