ExamplesΒΆ
If your features are collected into a features.csv file (each row is a subject, with features along the columns), running neuropredict is as simple as:
np_classify -m /project/meta_data.csv -d /project/features.csv
Where the meta_data.csv file specifies which subject belongs to which class. For example, if you have a dataset with the following three classes: 5 controls, 6 disease_one and 9 other_disease, all you would need to do is produce a meta data file as shown below (specifying a class label for each subject):
3071,controls
3069,controls
3064,controls
3063,controls
3057,controls
5004,disease_one
5074,disease_one
5077,disease_one
5001,disease_one
5002,disease_one
5003,disease_one
5000,other_disease
5006,other_disease
5013,other_disease
5014,other_disease
5016,other_disease
5018,other_disease
5019,other_disease
5021,other_disease
5022,other_disease
and neuropredict will produce the figures similar to the following:
The higher resolution PDFs are included in the docs folder.
In addition to the figures, the data underlying the figures is also exported in the form of multiple separate CSV files to the exported_results folder under the chosen output directory.
You can choose values for many other parameters, and specify the features in few other formats. For examples, when using pyradigm as in the input format and choosing train_perc to be 75% and the repeating the CV 250 times, the typical output on the command line would look something like :
neuropredict -y /project/*.MLDataset.pkl -m /project/meta_FourClasses.csv -o /project/predictions -t 0.75 -n 250
Requested features for analysis:
get_pyradigm from chebyshev.MLDataset.pkl
get_pyradigm from chebyshev_neg.MLDataset.pkl
get_pyradigm from chi_square.MLDataset.pkl
get_pyradigm from correlate_1.MLDataset.pkl
get_pyradigm from correlate.MLDataset.pkl
get_pyradigm from cosine_1.MLDataset.pkl
get_pyradigm from cosine_2.MLDataset.pkl
get_pyradigm from cosine_alt.MLDataset.pkl
get_pyradigm from cosine.MLDataset.pkl
get_pyradigm from euclidean.MLDataset.pkl
get_pyradigm from fidelity_based.MLDataset.pkl
Different classes in the training set are stratified to match the smallest class!
CV repetition 0
feature 0 weight_chebyshev : balanced accuracy: 0.3018
feature 1 weight_chebyshev_neg : balanced accuracy: 0.2917
feature 2 weight_chi_square : balanced accuracy: 0.2603
feature 3 weight_correlate_1 : balanced accuracy: 0.3271
feature 4 weight_correlate : balanced accuracy: 0.3647
feature 5 weight_cosine_1 : balanced accuracy: 0.3202
feature 6 weight_cosine_2 : balanced accuracy: 0.2869
feature 7 weight_cosine_alt : balanced accuracy: 0.3656
feature 8 weight_cosine : balanced accuracy: 0.3197
feature 9 weight_euclidean : balanced accuracy: 0.2579
feature 10 weight_fidelity_based : balanced accuracy: 0.1190
CV repetition 1
feature 0 weight_chebyshev : balanced accuracy: 0.3416
feature 1 weight_chebyshev_neg : balanced accuracy: 0.3761
feature 2 weight_chi_square : balanced accuracy: 0.3748
feature 3 weight_correlate_1 : balanced accuracy: 0.3397
feature 4 weight_correlate : balanced accuracy: 0.4087
feature 5 weight_cosine_1 : balanced accuracy: 0.3074
feature 6 weight_cosine_2 : balanced accuracy: 0.4059
feature 7 weight_cosine_alt : balanced accuracy: 0.3658
feature 8 weight_cosine : balanced accuracy: 0.3290
feature 9 weight_euclidean : balanced accuracy: 0.2662
feature 10 weight_fidelity_based : balanced accuracy: 0.2090
CV repetition 2
. . . .
. . . .
. . . .
CV repetition n
Note
pyradigm is a python class (data structure) aimed to ease your ML workflow - check it out here.