Compare different histogram2d binnings using the same edges-CodePudding

I have a dataset that looks like this:

tsne_results_x  tsne_results_y  team_id
0   -22.796648  -26.514051  107
1   11.985229   40.674446   107
2   -28.231720  -49.302216  107
3   31.942875   -14.427114  107
4   -46.436501  -7.750005   107
76  24.252718   -20.551889  8071
77  2.362172    17.170067   8071
78  7.212677    -9.056982   8071
79  -5.865472   -32.999077  8071

I want to bin the tsne_results_x and tsne_results_y columns and for that I am using numpy function histogram2d

grid, xe, ye = np.histogram2d(df['tsne_results_x'], df['tsne_results_y'], bins=15)
gridx = np.linspace(min(df['tsne_results_x']),max(df['tsne_results_x']),15)
gridy = np.linspace(min(df['tsne_results_y']),max(df['tsne_results_y']),15)

plt.figure()
#plt.plot(x, y, 'ro')
plt.grid(True)

#plt.figure()
plt.pcolormesh(gridx, gridy, grid)
plt.colorbar()

plt.show()

However, as you can see, I have a few team_ids in the data frame and I would like to compare one team's individual bins to the whole data frame. For example, for one team, at one specific bin, I want to divide it by the total count that includes all the teams.

So, I thought that running histogram2d on a specific team dataset, using the same linespace for the whole dataset would do the trick. It does not, because the histogram2d will bin the one_team_df differently because the data has different ranges

one_team_df = df.loc[(df['team_id'] == str(299))]

grid_team, a, b = np.histogram2d(one_team_df['tsne_results_x'], one_team_df['tsne_results_y'], bins=15)



gridx = np.linspace(min(df['tsne_results_x']),max(df['tsne_results_x']),15)
gridy = np.linspace(min(df['tsne_results_y']),max(df['tsne_results_y']),15)

plt.figure()
#plt.plot(x, y, 'ro')
plt.grid(True)

#plt.figure()
plt.pcolormesh(gridx, gridy, grid_team)
#plt.plot(x, y, 'ro')
plt.colorbar()

plt.show()

I would like to know how do I make these two representations comparable. Is it possible to run histogram2d giving the xedges and yedges ? This way I could bin one team using the edges of the overall binning.

Regards

CodePudding user response：

documentation of np.histomgram2d

binsint or array_like or [int, int] or [array, array], optional
The bin specification:

If int, the number of bins for the two dimensions (nx=ny=bins).

If array_like, the bin edges for the two dimensions (x_edges=y_edges=bins).

If [int, int], the number of bins in each dimension (nx, ny = bins).

If [array, array], the bin edges in each dimension (x_edges, y_edges = bins).

A combination [int, array] or [array, int], where int is the number of bins and array is the bin edges.

This means you can specify the bins as you want. For instance:

grid_team, a, b = np.histogram2d(
    one_team_df['tsne_results_x'], one_team_df['tsne_results_y'], 
    bins=[np.linspace(-40,40,15), np.linspace(-40,40,15)]
)