for 循环中的子图

Question

I want to create subplots within a loop, but the outcome is not excactly what I imagined. I want scatter plots in one big plot. The data originates from the matching columns of two dataframes DF and dfr . DF and dfr have the same amount of rows columns and indexes. The first two columns of both dataframes should be excluded.

This is my approach, but I get i plots with one subplot each. What am I missing?

        measurements = 9
        for i in range(2,measurements+1):
            try:
                x = DF.iloc[1:,i]
                y = dfr.iloc[1:,i]
        
                inds = ~np.logical_or(np.isnan(x), np.isnan(y))
                x = x[inds]
                y = y[inds]

                xy = np.vstack([x,y])
                z = gaussian_kde(xy)(xy)
                b, m = polyfit(x, y, 1)
                
                fig, ax = plt.subplots(measurements+1,facecolor='w', edgecolor='k')
                ax[i].scatter(x, y, c=z,  s=50, cmap='jet', edgecolor='', label=None, picker=True, zorder= 2)
                ax[i].plot(x, b + m * x, '-')
      
            except KeyError:
                continue
        plt.show()

Currently I get several plots, but i would like to have one with multipile subplots.

Answer 1

Indeed, you have to put fig, ax = plt.subplots() out of the loop.

A few other things:

• Setting edgecolor='' that way might raise an error. Remove it, or add a specific color.
• I am sure if using try and except KeyError is relevant in your code. Python raises a KeyError whenever a dict() object is requested (using the format a = adict[key] ) and the key is not in the dictionary. Maybe for: x = x[inds] ? if so, I would suggest do this check earlier in your process.

Try this:

measurements = 9

fig, ax = plt.subplots(measurements+1, facecolor='w', edgecolor='k')

for i in range(2, measurements+1):
    try:
        x = DF.iloc[1:,i]
        y = dfr.iloc[1:,i]

        inds = ~np.logical_or(np.isnan(x), np.isnan(y))
        x = x[inds]
        y = y[inds]

        xy = np.vstack([x,y])
        z = stats.gaussian_kde(xy)(xy)
        b, m = np.polyfit(x, y, 1)
        ax[i].scatter(x, y, c=z,  s=50, cmap='jet', label=None, picker=True, zorder= 2)
        ax[i].plot(x, b + m * x, '-')
    except KeyError:
        # Temporarily pass but ideally, do something
        pass

plt.show()