No project description provided
Project description
TimeSeriesRNNInterpretor
TimeSeriesRNNInterpretor, Interprets the RNN black box model. It internally uses shap.DeepExplainer. We have tried to interpret only single instance with respect to time. We have additionally included monthly effect of the instance.
- ✨TimeSeriesRNNInterpretor✨
Parameters
Parameters required to Pass
Parameter | To be passed | Format |
---|---|---|
model | RNN model | tensorflow V1 behaviour |
background | training data sample | Numpy array |
test | Instance data which we want to explain | Numpy array |
features | feature_names or column names | df.columns |
max_output | max output value(1) | 1 for binary classification |
max_display | max display features for waterfall plot | a number < no of features |
decision_range | max display features for decision plot | a number < no of features |
Measure | "absolute" or "relative" or "total " | for plotting waterfall plot w.r.t basevalue |
dependence_feature_3 | 3rd feature name to plot dependence plot | a feature name or index of feature |
dependence_feature_2 | 2nd feature name to plot dependence plot | a feature name or index of the feature |
Calling
from TimeSeriesRNNInterpretor import TimeSeriesRNNInterpretor
temp = TimeSeriesInterpretorRNN(model=model,background=background,test=test,features=features,max_output=1,max_display=15,decision_range=30,Measure="absolute",dependence_feature_3="age.",dependence_feature_2="gender")
Plotting Different Plots
temp.global_patient() # To the know the feature importance of a instance
temp.monthly_chart() # timeframe charts of a instance (day,month, or any time frame)
temp.two_month_compare(17, 18) # Specify two timeframe numbers of a instancee to compare (days,months, or any time frames)
temp.single_month(18) # Specify the required specific timeframe number of instance(day,month, or any time frame)
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Close
Hashes for TimeSeriesInterpretorRNN-0.0.11.tar.gz
Algorithm | Hash digest | |
---|---|---|
SHA256 | f0335001e43343bfd0ee24bd44f15e0efe6bfc7c4a122c72557f6062d782efa1 |
|
MD5 | 703b3853bc9a4af6f0849038f1d5fa2c |
|
BLAKE2b-256 | 2fe636b5c277a4f17cb63ccafb7f69f261646ca778448c4ca79d7a3e73d2fe48 |
Close
Hashes for TimeSeriesInterpretorRNN-0.0.11-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 170bb6db489a8400bb2f8dbff09997f9e6dcb9089da0b56daaf84c8a88f984e2 |
|
MD5 | 3f2a9199fe4d5d8e11a2cfd6bc60af58 |
|
BLAKE2b-256 | fddba954da0772f734710f772699a4854a3c761d2be37bad7b79ad6f02bce727 |