0xideas/hyperframe

HyperFrame

The aim of this project is to provide a high-dimensional analogue to the two-dimensional pandas DataFrame.

This allows its user to organise information where the interaction of several factors is of interest.

The HyperFrame allows for the easy setting and saving of data for storage, and the fast, interactive creation of two-dimensional pandas DataFrames of any combination of two factors for data exploration.

import numpy as np
import pandas as pd
from sklearn.neighbors import KNeighborsClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import RandomForestClassifier
from hyperframe import HyperFrame
from sklearn.model_selection import train_test_split
from demo.helpers import metrics, X, y

X_train, X_test, y_train, y_test = \
    train_test_split(X, y, test_size=0.33, random_state=42)

clf = KNeighborsClassifier()
clf.fit(X_train, y_train)

KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',
                     metric_params=None, n_jobs=None, n_neighbors=5, p=2,
                     weights='uniform')

Initialisation

dimension_labels = ["train_test", "species", "metric"]

index_labels = {"train_test": ["train", "test"],
                "species": ["setosa", "versicolor", "virginica"],
                "metric": ["precision", "recall", "f1"]}

scores = HyperFrame(dimension_labels, index_labels)

Setting data

yhat = clf.predict(X_train)
#iset alternative 1
scores.iset(metrics(y_train, yhat), "train", "", "")

<hyperframe.HyperFrame at 0x7ff4d4241320>

yhat = clf.predict(X_test)
#iset alternative 2
scores.iset(metrics(y_test, yhat), train_test="test")

<hyperframe.HyperFrame at 0x7ff4d4241320>

Getting data

#iget alternative 1
scores.iget("train", "", "", return_type="pandas").round(2)

	precision	recall	f1
setosa	0.89	1.00	0.94
versicolor	0.71	0.71	0.71
virginica	0.80	0.71	0.75

#iget alternative 2
scores.iget(species="versicolor", return_type="pandas").round(2)

	precision	recall	f1
train	0.71	0.71	0.71
test	0.70	0.47	0.56

#iget alternative 3
scores.iget0("species", "train_test", return_type="pandas").round(2)

{'metric': 'precision'}

	setosa	versicolor	virginica
train	0.89	0.71	0.80
test	0.95	0.70	0.65

Initialising a second HyperFrame

scores_lr = HyperFrame(dimension_labels, index_labels)
clf = LogisticRegression(penalty="none", max_iter=1000)
clf.fit(X_train, y_train)

yhat = clf.predict(X_train)
scores_lr.iset(metrics(y_train, yhat), "train", "", "")

yhat = clf.predict(X_test)
scores_lr.iset(metrics(y_test, yhat), "test", "", "")

<hyperframe.HyperFrame at 0x7ff4d4231588>

Merging

print("scores shape: {}".format(scores.shape))
print("scores_lr shape: {}".format(scores_lr.shape))

scores shape: (2, 3, 3)
scores_lr shape: (2, 3, 3)

scores_models = scores.merge(scores_lr, "model", ["knn", "logistic regression"])

scores_models.iget("test", "", "f1", "", return_type="pandas").round(2)

	knn	logistic regression
setosa	0.97	0.95
versicolor	0.56	0.58
virginica	0.72	0.72

scores_models.iget("", "", "f1", "logistic regression", return_type="pandas").round(2)

	setosa	versicolor	virginica
train	0.92	0.74	0.78
test	0.95	0.58	0.72

Initialising a third HyperFrame

scores_rf = HyperFrame(dimension_labels, index_labels)
clf = RandomForestClassifier()
clf.fit(X_train, y_train)

yhat = clf.predict(X_train)
scores_rf.iset(metrics(y_train, yhat), "train", "", "")

yhat = clf.predict(X_test)
scores_rf.iset(metrics(y_test, yhat), "test", "", "")

<hyperframe.HyperFrame at 0x7ff4d41cd978>

scores_rf.iget("test", "", "", return_type="pandas").round(2)

	precision	recall	f1
setosa	0.95	0.95	0.95
versicolor	0.75	0.40	0.52
virginica	0.61	0.88	0.72

Expanding A DataFrame

print("scores_models shape: {}".format(scores_models.shape))
print("scores_rf shape: {}".format(scores_rf.shape))

scores_models shape: (2, 3, 3, 2)
scores_rf shape: (2, 3, 3)

scores_models = scores_models.expand(scores_rf, "model", "random forest")

scores_models.iget("test", "", "f1", "", return_type="pandas").round(2)

	knn	logistic regression	random forest
setosa	0.97	0.95	0.95
versicolor	0.56	0.58	0.52
virginica	0.72	0.72	0.72

Simple Mathematical Operations

scores.max("train_test").iget("", "", return_type="pandas")

	precision	recall	f1
setosa	0.950000	1.000000	0.974359
versicolor	0.714286	0.714286	0.714286
virginica	0.800000	0.812500	0.750000

scores.min("train_test", "metric").iget("", return_type="pandas")

setosa        0.885714
versicolor    0.466667
virginica     0.650000
dtype: float64

scores.mean("train_test", "species", "metric")

0.7810886435641339

scores.sum()

14.059595584154408

Writing to file

scores_models.write_file("./demo/scores_models")

Reading from file

scores_models = scores_models.read_file("./demo/scores_models")