plot - Seaborn tsplot windowed estimators -
i want tsplot estimator windowed function rolling mean rather mean. ideally i'd pass rolling mean function estimator parameter of tsplot() , individual timepoints passed estimator. so, looks i'm stuck pre-processing data. is correct? there approach i'm overlooking here? i don't think quite understand you're trying do, bootstrap function used in tsplot compute confidence interval gets whole array , axis=0 , , resamples rows of array before reducing on operations. seems work: import numpy np import pandas pd import seaborn sns import matplotlib.pyplot plt data = np.cumsum(np.random.randn(25, 40), axis=1) sns.tsplot(data=data) def rolling_mean(data, axis=0): return pd.rolling_mean(data, 4, axis=1).mean(axis=axis) sns.tsplot(data=data, estimator=rolling_mean)