Add dvc yaml solution lab10

.gitignore

@@ -1 +1,4 @@
 /imdb_movies.csv
+/train.csv
+/test.csv
+/results.csv

dvc.lock

@@ -0,0 +1,34 @@
+schema: '2.0'
+stages:
+  split:
+    cmd: python lab_10_prepare.py
+    deps:
+    - path: imdb_movies.csv
+      md5: cf6471460161d4e0a85271c467845d7c
+      size: 50492646
+    - path: lab_10_prepare.py
+      md5: e0f6e525730ab3d991b5e5777ffa2ae0
+      size: 1324
+    outs:
+    - path: test.csv
+      md5: 9bd42fac150dd8a33d32b6326921d984
+      size: 68005
+    - path: train.csv
+      md5: 7ba1b2b4673781406812f35569cb1ed0
+      size: 204232
+  train:
+    cmd: python3 lab_10_train.py
+    deps:
+    - path: lab_10_train.py
+      md5: 7717f393a6f1c6aea2b145ea1f2f6dd3
+      size: 1285
+    - path: test.csv
+      md5: 9bd42fac150dd8a33d32b6326921d984
+      size: 68005
+    - path: train.csv
+      md5: 7ba1b2b4673781406812f35569cb1ed0
+      size: 204232
+    outs:
+    - path: results.csv
+      md5: a52750b686aaeadd7cf4436cbe6904b5
+      size: 16046

dvc.yaml

@@ -0,0 +1,17 @@
+stages:
+  split:
+    cmd: python lab_10_prepare.py
+    deps:
+    - imdb_movies.csv
+    - lab_10_prepare.py
+    outs:
+    - test.csv
+    - train.csv
+  train:
+    cmd: python3 lab_10_train.py
+    deps:
+    - lab_10_train.py
+    - test.csv
+    - train.csv
+    outs:
+    - results.csv

lab_10_prepare.py

@@ -0,0 +1,50 @@
+"""
+Download dataset between 10-20 mb,
+Split it into train/dev/test
+Return dataset info (length, max, min etc.)
+"""
+
+import string
+
+import pandas as pd
+from sklearn import preprocessing
+from sklearn.model_selection import train_test_split
+
+movies_data = pd.read_csv("imdb_movies.csv")
+
+# Drop rows with missing values
+movies_data.dropna(inplace=True)
+
+# Remove not interesting columns
+drop_columns = ["title_id", "certificate", "title", "plot"]
+drop_columns2 = [
+    "original_title",
+    "countries",
+    "genres",
+    "director",
+    "cast",
+    "release_date",
+]
+drop_columns = drop_columns + drop_columns2
+
+movies_data.drop(labels=drop_columns, axis=1, inplace=True)
+
+
+# Remove ',' from votes number and change type to int
+movies_data["votes_number"] = (movies_data["votes_number"].str.replace(",", "")).astype(
+    int
+)
+
+# Normalize number values
+scaler = preprocessing.MinMaxScaler()
+movies_data[["votes_number", "year", "runtime"]] = scaler.fit_transform(
+    movies_data[["votes_number", "year", "runtime"]]
+)
+
+# Split set to train/dev/test 6:2:2 ratio and save to .csv file
+train, dev = train_test_split(movies_data, train_size=0.6, test_size=0.4, shuffle=True)
+dev, test = train_test_split(dev, train_size=0.5, test_size=0.5, shuffle=True)
+
+train.to_csv("train.csv")
+dev.to_csv("dev.csv")
+test.to_csv("test.csv")

lab_10_train.py

@@ -0,0 +1,48 @@
+import pandas as pd
+from sklearn.metrics import mean_absolute_error
+from tensorflow.keras.callbacks import EarlyStopping
+from tensorflow.keras.layers import Dense, Dropout
+from tensorflow.keras.models import Sequential
+
+movies_train = pd.read_csv("train.csv")
+X_train = movies_train.drop("rating", axis=1)
+Y_train = movies_train["rating"]
+
+movies_test = pd.read_csv("test.csv")
+X_test = movies_test.drop("rating", axis=1)
+Y_test = movies_test["rating"]
+
+# Set up model
+model = Sequential()
+model.add(Dense(8, activation="relu"))
+model.add(Dropout(0.5))
+model.add(Dense(3, activation="relu"))
+model.add(Dropout(0.5))
+model.add(Dense(1))
+model.compile(optimizer="adam", loss="mse")
+
+early_stop = EarlyStopping(monitor="val_loss", mode="min", verbose=1, patience=10)
+
+model.fit(
+    x=X_train,
+    y=Y_train.values,
+    validation_data=(X_test, Y_test.values),
+    batch_size=128,
+    epochs=400,
+    callbacks=[early_stop],
+)
+
+# Predict movie ratings
+predictions = model.predict(X_test)
+
+pd.DataFrame(predictions).to_csv("results.csv")
+
+
+# Compare outputs
+for i, score in enumerate(predictions):
+    print(f"Original score: {Y_test.iloc[i]} Predicted score: {score} \n")
+    print(f"Difference is : {Y_test.iloc[i] - score}")
+
+
+# Evaluate
+print(mean_absolute_error(Y_test, predictions))

requirements.txt

@@ -11,4 +11,5 @@ gast==0.3.3
 sacred==0.8.2
 GitPython==3.1.14
 matplotlib==3.3.4
-mlflow==1.17.0
+mlflow==1.17.0
+dvc==2.3.0