I have a neural network that predicts frequencies for four different features in my dataset. I then test my network and therefore for each input I get a vector of size four where each entry corresponds to frequency of a given feature. Now I want to plot a 2D density plot of true frequencies vs predicted frequencies.
true: [0.000345,0.99183,0,0] prediction: [0.0212,0.738,0.004,0.006]
true: [0,0.9937,0,0.00013] prediction: [0.005,0.983,0.04,0.01]
I basically plot 2D density plot for each of the features separately. That is I would plot true vs predictions for feature # 1, then for feature # 2 and so on.
import seaborn as sns
sns.jointplot(x=feature_1_true, y=feature_1_pred, kind='scatter')
sns.jointplot(x=feature_2_true, y=feature_2_pred, kind='scatter')
My question is there a way to visualise such a vector in totality on a density plot instead of plotting features separately. If you had a vector of such kind how would you plot it's density plot and how would you plot true vs predictions.