fuzzy-rules-generator/temp_viscosity_regression.ipynb

In [2]:

import pandas as pd

train = pd.read_csv("data/viscosity_train.csv", sep=";", decimal=",")
test = pd.read_csv("data/viscosity_test.csv", sep=";", decimal=",")

display(train.head())
display(test.head())

	T	Al2O3	TiO2	Viscosity
0	20	0.0	0.0	3.707
1	25	0.0	0.0	3.180
2	35	0.0	0.0	2.361
3	45	0.0	0.0	1.832
4	50	0.0	0.0	1.629

	T	Al2O3	TiO2	Viscosity
0	30	0.00	0.0	2.716
1	40	0.00	0.0	2.073
2	60	0.00	0.0	1.329
3	65	0.00	0.0	1.211
4	25	0.05	0.0	4.120

In [3]:

y_train = train["T"]
X_train = train.drop(["T"], axis=1)

display(X_train.head())
display(y_train.head())

y_test = test["T"]
X_test = test.drop(["T"], axis=1)

display(X_test.head())
display(y_test.head())

	Al2O3	TiO2	Viscosity
0	0.0	0.0	3.707
1	0.0	0.0	3.180
2	0.0	0.0	2.361
3	0.0	0.0	1.832
4	0.0	0.0	1.629

0    20
1    25
2    35
3    45
4    50
Name: T, dtype: int64

	Al2O3	TiO2	Viscosity
0	0.00	0.0	2.716
1	0.00	0.0	2.073
2	0.00	0.0	1.329
3	0.00	0.0	1.211
4	0.05	0.0	4.120

0    30
1    40
2    60
3    65
4    25
Name: T, dtype: int64

In [10]:

from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import PolynomialFeatures
from sklearn import linear_model, tree, neighbors, ensemble

random_state = 9

models = {
    "linear": {"model": linear_model.LinearRegression(n_jobs=-1)},
    "linear_poly": {
        "model": make_pipeline(
            PolynomialFeatures(degree=2),
            linear_model.LinearRegression(fit_intercept=False, n_jobs=-1),
        )
    },
    "linear_interact": {
        "model": make_pipeline(
            PolynomialFeatures(interaction_only=True),
            linear_model.LinearRegression(fit_intercept=False, n_jobs=-1),
        )
    },
    "ridge": {"model": linear_model.RidgeCV()},
    "decision_tree": {
        "model": tree.DecisionTreeRegressor(random_state=random_state, max_depth=6, criterion="absolute_error")
    },
    "knn": {"model": neighbors.KNeighborsRegressor(n_neighbors=7, n_jobs=-1)},
    "random_forest": {
        "model": ensemble.RandomForestRegressor(
            max_depth=7, random_state=random_state, n_jobs=-1
        )
    },
}

In [11]:

import math
from sklearn import metrics

for model_name in models.keys():
    print(f"Model: {model_name}")
    fitted_model = models[model_name]["model"].fit(
        X_train.values, y_train.values.ravel()
    )
    y_train_pred = fitted_model.predict(X_train.values)
    y_test_pred = fitted_model.predict(X_test.values)
    models[model_name]["fitted"] = fitted_model
    models[model_name]["MSE_train"] = metrics.mean_squared_error(y_train, y_train_pred)
    models[model_name]["MSE_test"] = metrics.mean_squared_error(y_test, y_test_pred)
    models[model_name]["MAE_train"] = metrics.mean_absolute_error(y_train, y_train_pred)
    models[model_name]["MAE_test"] = metrics.mean_absolute_error(y_test, y_test_pred)
    models[model_name]["R2_train"] = metrics.r2_score(y_train, y_train_pred)
    models[model_name]["R2_test"] = metrics.r2_score(y_test, y_test_pred)

Model: linear
Model: linear_poly
Model: linear_interact
Model: ridge
Model: decision_tree
Model: knn
Model: random_forest

In [12]:

reg_metrics = pd.DataFrame.from_dict(models, "index")[
    ["MSE_train", "MSE_test", "MAE_train", "MAE_test", "R2_train", "R2_test"]
]
reg_metrics.sort_values(by="MAE_test").style.background_gradient(
    cmap="viridis", low=1, high=0.3, subset=["MSE_train", "MSE_test"]
).background_gradient(cmap="plasma", low=0.3, high=1, subset=["MAE_test", "R2_test"])

Out[12]:

	MSE_train	MSE_test	MAE_train	MAE_test	R2_train	R2_test
linear_poly	4.827768	4.877296	1.522643	1.743058	0.980864	0.974964
linear_interact	21.786348	23.459572	3.830996	4.381115	0.913644	0.879577
linear	27.766510	35.430313	4.088006	5.106782	0.889940	0.818129
ridge	31.827476	36.230606	4.383008	5.226480	0.873843	0.814021
random_forest	8.525285	45.444651	2.542935	5.749510	0.966208	0.766723
decision_tree	3.289474	45.588235	0.921053	6.176471	0.986961	0.765986
knn	61.855532	64.165666	6.522556	6.806723	0.754819	0.670624

In [13]:

model = models["decision_tree"]["fitted"]
rules = tree.export_text(
    model, feature_names=X_train.columns.values.tolist()
)
print(rules)

|--- Viscosity <= 2.86
|   |--- Viscosity <= 1.38
|   |   |--- value: [70.00]
|   |--- Viscosity >  1.38
|   |   |--- Viscosity <= 1.78
|   |   |   |--- Viscosity <= 1.72
|   |   |   |   |--- TiO2 <= 0.03
|   |   |   |   |   |--- Al2O3 <= 0.03
|   |   |   |   |   |   |--- value: [52.50]
|   |   |   |   |   |--- Al2O3 >  0.03
|   |   |   |   |   |   |--- value: [57.50]
|   |   |   |   |--- TiO2 >  0.03
|   |   |   |   |   |--- value: [60.00]
|   |   |   |--- Viscosity >  1.72
|   |   |   |   |--- value: [70.00]
|   |   |--- Viscosity >  1.78
|   |   |   |--- TiO2 <= 0.18
|   |   |   |   |--- Al2O3 <= 0.18
|   |   |   |   |   |--- Viscosity <= 2.24
|   |   |   |   |   |   |--- value: [50.00]
|   |   |   |   |   |--- Viscosity >  2.24
|   |   |   |   |   |   |--- value: [40.00]
|   |   |   |   |--- Al2O3 >  0.18
|   |   |   |   |   |--- Viscosity <= 2.29
|   |   |   |   |   |   |--- value: [60.00]
|   |   |   |   |   |--- Viscosity >  2.29
|   |   |   |   |   |   |--- value: [55.00]
|   |   |   |--- TiO2 >  0.18
|   |   |   |   |--- Viscosity <= 2.22
|   |   |   |   |   |--- value: [70.00]
|   |   |   |   |--- Viscosity >  2.22
|   |   |   |   |   |--- Viscosity <= 2.69
|   |   |   |   |   |   |--- value: [60.00]
|   |   |   |   |   |--- Viscosity >  2.69
|   |   |   |   |   |   |--- value: [55.00]
|--- Viscosity >  2.86
|   |--- Viscosity <= 3.64
|   |   |--- TiO2 <= 0.18
|   |   |   |--- Viscosity <= 3.15
|   |   |   |   |--- value: [35.00]
|   |   |   |--- Viscosity >  3.15
|   |   |   |   |--- Viscosity <= 3.47
|   |   |   |   |   |--- Al2O3 <= 0.03
|   |   |   |   |   |   |--- value: [25.00]
|   |   |   |   |   |--- Al2O3 >  0.03
|   |   |   |   |   |   |--- value: [30.00]
|   |   |   |   |--- Viscosity >  3.47
|   |   |   |   |   |--- value: [40.00]
|   |   |--- TiO2 >  0.18
|   |   |   |--- value: [45.00]
|   |--- Viscosity >  3.64
|   |   |--- Viscosity <= 6.27
|   |   |   |--- TiO2 <= 0.18
|   |   |   |   |--- Al2O3 <= 0.18
|   |   |   |   |   |--- TiO2 <= 0.03
|   |   |   |   |   |   |--- value: [20.00]
|   |   |   |   |   |--- TiO2 >  0.03
|   |   |   |   |   |   |--- value: [22.50]
|   |   |   |   |--- Al2O3 >  0.18
|   |   |   |   |   |--- Viscosity <= 4.42
|   |   |   |   |   |   |--- value: [35.00]
|   |   |   |   |   |--- Viscosity >  4.42
|   |   |   |   |   |   |--- value: [27.50]
|   |   |   |--- TiO2 >  0.18
|   |   |   |   |--- Viscosity <= 4.65
|   |   |   |   |   |--- value: [35.00]
|   |   |   |   |--- Viscosity >  4.65
|   |   |   |   |   |--- Viscosity <= 5.40
|   |   |   |   |   |   |--- value: [30.00]
|   |   |   |   |   |--- Viscosity >  5.40
|   |   |   |   |   |   |--- value: [25.00]
|   |   |--- Viscosity >  6.27
|   |   |   |--- value: [20.00]

In [14]:

import pickle

pickle.dump(model, open("data/temp_viscosity_tree.model.sav", "wb"))