0

I was wondering how I can adjust my dates here so that they're smaller and fit better on the output.I've tried rotating but they just seem like they're floating aimlessly below the chart. Also Id like to make the legend just either have y_pred and y_test one time somewhere where it doesn't overlap my charts.

These subplots are added via a loop and it wont always be the same number of loops. For reference no_splits will determines how many loops are run via the TimeSeriesSplit method. Ive removed a lot of the irrelevant code so it's easier to follow

Here is my code:

fig = plt.figure()
    tscv = TimeSeriesSplit(n_splits=self.no_splits)
    for train_index, test_index in tqdm(tscv.split(X)):
        X_train, X_test = X.iloc[train_index], X.iloc[test_index]
        y_train, y_test = y.iloc[train_index], y.iloc[test_index]

        # predict y values
        y_pred = self.regressor.predict(X_test)


        # plot y_pred vs y_test
        y_df = pd.DataFrame(index= X_test_index)
        y_pred = y_pred.reshape(len(y_pred), )
        y_test = y_test.reshape(len(y_test), )
        y_df['y_pred'] = y_pred
        y_df['y_test'] = y_test

        ax = fig.add_subplot(int(sqrt(self.no_splits)), int(sqrt(self.no_splits)+1), i)


        y_df.plot(title = 'Split{}'.format(i), ax=ax, legend=False)
        ax.tick_params(axis='x', rotation=45)

        plt.figlegend()
    plt.subplots_adjust(wspace=0, hspace=0)
    plt.show()

enter image description here

enter image description here

novawaly
  • 1,051
  • 3
  • 12
  • 27
  • How about using `plt.tight_layout()` before `plt.show()` and commenting out `plt.subplots_adjust(wspace=0, hspace=0)`? – Sheldore Jun 13 '19 at 11:35
  • so that shrinks the charts down to fit the dates - id like to shrink the datess to fit the charts and maybe have the dates line up better with the hash marks? – novawaly Jun 13 '19 at 11:41
  • "Shrinking the dates to fit the charts" would mean to use a fontsize of 2 or so, which will be unreadable. Are you sure you don't want more space between the subplots? – ImportanceOfBeingErnest Jun 13 '19 at 11:54
  • I just made an edit to show what happens when i use tight_layout with more charts. I want more space but not at the expense of my charts – novawaly Jun 13 '19 at 11:57

1 Answers1

1

Regarding the date date-labels: You can assign the tick-alignment in the rotation command as shown in this post.

To shrink the labels you have two options:

Option A: Import matplotlib.dates to get access to DateFormatter and choose a format that results in smaller labels. (eg. omit the year or something). You can then also use locators to space the labels out differently.

Option B: use the rc_params or tick_params to define fontsize, family, etc. This post should get you started.

As you see, there should be a ot of material online to get you going...

Regarding the legend

You can set a plot to have no legend entry using plt.plot(x, y, label='_nolabel'). This you could combine with a for loop for example, to only draw labels on the first iteration.

for i, (train_index, test_index) in enumerate(tqdm(tscv.split(X))):
    if i==0:
        plt.plot(x, y, label=label)
    else:
        plt.plot(x, y, label='_nolabel')
flurble
  • 1,086
  • 7
  • 21