How to transfer pandas .plot() to Matplotlib .errorbar()

Question

I'm looking to plot error bars on a line plot I did using pandas's .plot() function

scores_xgb.plot(x='size', y='MSE_mean_tot', kind='line',logx=True,title='XGBoost 5 samples, 5 fold CV')

Running this gives me the following plot:

For plotting the error bars, I choose to use .errorbar() from Matplotlib. When running the cell

plt.errorbar(x=scores_xgb.size, y=scores_xgb.MSE_mean_tot, yerr=std_xgb ,title='XGBoost 5 samples, 5 fold CV')
plt.show()

I receive the following error message:

ValueError: 'x' and 'y' must have the same size

This confuses me, as I use the same Dataframe in both examples, each time using the same variable for x and y respectively, therefore it has the same size (12) both times.

NB: the yerr = std_xgb also has size 12.

Answer 1

There is a property on pandas.DataFrame objects named size , and it's a number, equal to the number of cells in the DataFrame (the product of the values in df.shape ). You're trying to access a column named size , but pandas chooses the property named size before it chooses the column name size . Since the single number has a shape of 1 but the columns in the dataframe have a length of 12, you're getting a shape mismatch.

Instead, use strings to index the dataframe and get the the columns:

plt.errorbar(x=scores_xgb['size'], y=scores_xgb['MSE_mean_tot'], yerr=std_xgb, title='XGBoost 5 samples, 5 fold CV')