Zadanie frameworki ML

Dockerfile

@@ -7,10 +7,13 @@ RUN apt install -y curl
 RUN pip3 install --user pandas
 RUN pip3 install --user matplotlib
 RUN pip3 install --user numpy
+RUN pip3 install --user tensorflow
+RUN pip3 install --user sklearn
 
 WORKDIR /app
 
 COPY ./preparations.sh ./
 COPY ./preprocesing_python.py ./
+COPY ./training.py ./
 
 # CMD ./preparations.sh

training.py

@@ -3,60 +3,87 @@ import pandas as pd
 import numpy as np
 import matplotlib.pyplot as plt
 import os
-import tensorflow as tf
 from countries_map import countries
+import tensorflow as tf
+from tensorflow.keras.layers import Input, Dense, Activation, Dropout
+from tensorflow.keras.models import Model
+from tensorflow.keras.callbacks import EarlyStopping
+from keras.models import Sequential
+from sklearn.metrics import mean_squared_error
+from tensorflow import keras
+from tensorflow.keras import layers
+from tensorflow.keras.layers.experimental import preprocessing
 
 
-def mapSet(set):
-    age = {"5-14 years": 0, "15-24 years": 1, "25-34 years": 2,
-           "35-54 years": 3, "55-74 years": 4, "75+ years": 5}
-    sex = {"male": 0, "female": 1}
-
-    set["age"].replace(age, inplace=True)
-    set["sex"].replace(sex, inplace=True)
-    set["country"].replace({v: k for k, v in countries.items()}, inplace=True)
-
-    return set
-
-
-column_names = ["country", "year", "sex", "age", "suicides_no", "population"]
-feature_names = ["country", "year", "sex", "age", "population"]
-label_name = column_names[4]
+age = {"5-14 years": 0, "15-24 years": 1, "25-34 years": 2,
+       "35-54 years": 3, "55-74 years": 4, "75+ years": 5}
+sex = {"male": 0, "female": 1}
 
+# wczytanie danych
 sc = pd.read_csv('who_suicide_statistics.csv')
+print(sc.shape)
 
+# Usunięcie niepełnych danych
+sc.dropna(inplace=True)
+
+# Kategoryzacja
+sc = pd.get_dummies(
+    sc, columns=['age', 'sex', 'country'], prefix='', prefix_sep='')
+
+
+# podział na train validate i test
 train, validate, test = np.split(sc.sample(frac=1, random_state=42),
                                  [int(.6*len(sc)), int(.8*len(sc))])
-train.dropna(inplace=True)
-validate.dropna(inplace=True)
-test.dropna(inplace=True)
 
-train_n = mapSet(train)
-validate_n = mapSet(validate)
-test_n = mapSet(validate)
+# podział train set
+X_train = train.loc[:, train.columns != 'suicides_no']
+y_train = train[['suicides_no']]
+X_test = test.loc[:, train.columns != 'suicides_no']
+y_test = test[['suicides_no']]
 
-train_csv = pd.DataFrame.to_csv(train_n, index=False)
+normalizer = preprocessing.Normalization()
+normalizer.adapt(np.array(X_train))
 
-train_dataset = tf.data.experimental.make_csv_dataset(
-    train_csv,
-    1000,
-    column_names=column_names,
-    label_name=label_name,
-    num_epochs=1)
+first = np.array(X_train[:1])
+with np.printoptions(precision=2, suppress=True):
+    print('First example:', first)
+    print()
+    print('Normalized:', normalizer(first).numpy())
 
-features, labels = next(iter(train_dataset))
-print(features)
+model = tf.keras.Sequential([
+    normalizer,
+    layers.Dense(units=1)
+])
+model.predict(X_train[:10])
 
-plt.scatter(features['year'],
-            features['age'],
-            c=labels,
-            cmap='sex')
+# Compile model
+model.compile(
+    optimizer=tf.optimizers.Adam(learning_rate=0.1),
+    loss='mean_absolute_error')
 
-plt.xlabel("year")
-plt.ylabel("age")
-plt.show()
+# Train model
+history = model.fit(
+    X_train, y_train,
+    epochs=30,
+    validation_split=0.2)
 
-print("Features: {}".format(feature_names))
-print("Label: {}".format(label_name))
+test_results = {}
 
-# print(train)
+test_results['model'] = model.evaluate(
+    X_test, y_test, verbose=0)
+
+test_predictions = model.predict(X_test).flatten()
+
+# a = plt.axes(aspect='equal')
+# plt.scatter(y_test, test_predictions)
+# plt.xlabel('True values [sucides_no]')
+# plt.ylabel('Predictions values [sucides_no]')
+# lims = [0, 5000]
+# plt.xlim(lims)
+# plt.ylim(lims)
+# _ = plt.plot(lims, lims)
+
+# plt.show()
+
+predictions = model.predict(X_test)
+pd.DataFrame(predictions).to_csv('results.csv')