"
],
"text/plain": [
" Speed Direction Temperature Power\n",
"780 NaN NaN NaN NaN"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"\n",
"\n",
"\n",
"Dataset 2 ~ Example of line where all values are missing\n"
]
},
{
"data": {
"text/html": [
"
\n",
"\n",
"
\n",
" \n",
"
\n",
"
\n",
"
Speed
\n",
"
Direction
\n",
"
Temperature
\n",
"
Power
\n",
"
\n",
" \n",
" \n",
"
\n",
"
902
\n",
"
NaN
\n",
"
NaN
\n",
"
NaN
\n",
"
NaN
\n",
"
\n",
" \n",
"
\n",
"
"
],
"text/plain": [
" Speed Direction Temperature Power\n",
"902 NaN NaN NaN NaN"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"f = lambda x : x[x.isna().all(axis=1)].head(1) \n",
"display_two(f,\"Example of line where all values are missing\",data1,data2)"
]
},
{
"cell_type": "markdown",
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-10T11:11:14.434751Z",
"start_time": "2021-10-10T11:11:14.430840Z"
}
},
"source": [
"We remove these kind of lines."
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:17.607001Z",
"start_time": "2021-10-17T21:24:17.597709Z"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Dataset 1: (6816, 4) \tDataset 2: (6934, 4)\n"
]
}
],
"source": [
"def remove_empty_lines(data) :\n",
" data = data.dropna(how='all')\n",
" data.reset_index(inplace=True, drop=True)\n",
" return data\n",
" \n",
"data1 = remove_empty_lines(data1)\n",
"data2 = remove_empty_lines(data2)\n",
"\n",
"print(\"Dataset 1:\", data1.shape, \"\\tDataset 2:\",data2.shape)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Lines with missing values\n",
"\n",
"Now we check the lines where there is at least one missing value."
]
},
{
"cell_type": "code",
"execution_count": 14,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:17.620036Z",
"start_time": "2021-10-17T21:24:17.608912Z"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Dataset 1 ~ Proportion of missing values (by line)\n",
"Speed 0.000000\n",
"Direction 0.000147\n",
"Temperature 0.000000\n",
"Power 0.000000\n",
"dtype: float64\n",
"\n",
"\n",
"\n",
"Dataset 2 ~ Proportion of missing values (by line)\n",
"Speed 0.007263\n",
"Direction 0.000144\n",
"Temperature 0.000000\n",
"Power 0.000000\n",
"dtype: float64\n"
]
}
],
"source": [
"f = lambda x : x.isna().sum(axis=0)/x.count()\n",
"print_two(f, \"Proportion of missing values (by line)\", data1, data2)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"For each column, the rows with at least one missing value correspond to less than 0.8% of the data. We will remove these lines (because it will not really change the distribution of our data)."
]
},
{
"cell_type": "code",
"execution_count": 15,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:17.637942Z",
"start_time": "2021-10-17T21:24:17.622295Z"
},
"scrolled": false
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Dataset 1 ~ Shape before removing missing values:\n",
"(6816, 4)\n",
"\n",
"\n",
"\n",
"Dataset 2 ~ Shape before removing missing values:\n",
"(6934, 4)\n",
"\n",
"\n",
"\n",
"\n",
"Dataset 1 ~ Shape after removing missing values:\n",
"(6815, 4)\n",
"\n",
"\n",
"\n",
"Dataset 2 ~ Shape after removing missing values:\n",
"(6883, 4)\n"
]
}
],
"source": [
"def remove_missing(data) :\n",
" data = data.dropna(how='any')\n",
" data.reset_index(inplace=True, drop=True)\n",
" return data\n",
"\n",
"f = lambda x : x.shape\n",
"print_two(f ,\"Shape before removing missing values:\",data1,data2)\n",
"\n",
"data1 = remove_missing(data1)\n",
"data2 = remove_missing(data2)\n",
"print(\"\\n\\n\\n\")\n",
"print_two(f,\"Shape after removing missing values:\",data1,data2)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## What about negative power"
]
},
{
"cell_type": "markdown",
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-10T13:19:13.947603Z",
"start_time": "2021-10-10T13:19:13.944262Z"
}
},
"source": [
"We could see that there were data with negative electrical powers. On peut afficher leur proportion :"
]
},
{
"cell_type": "code",
"execution_count": 16,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:17.651878Z",
"start_time": "2021-10-17T21:24:17.640106Z"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Dataset 1 ~ Proportion of negative electrical power\n",
"0.10975788701393983\n",
"\n",
"\n",
"\n",
"Dataset 2 ~ Proportion of negative electrical power\n",
"0.16998401859654222\n"
]
}
],
"source": [
"f = lambda x : ( x[x.Power < 0].count()/x.count() ).Power\n",
"print_two(f,\"Proportion of negative electrical power\", data1, data2)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"This corresponds to about 10 to 20% of the data. This number is not negligible. We can ask ourselves if these negative values make sense. After research and reflection, we have concluded that they do. Indeed, it is possible that a wind turbine consumes more electricity than it produces (in very low wind conditions). \n",
"\n",
"So we will not apply any particular pre-treatment for the lines where the power is negative. Especially since the power is the label to predict. If it is negative, it must be taken into account for the training of the model."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Standardization\n",
"\n",
"Now we standardize our features in order to put each feature on the same scale. We will center-reduce each value with the following formula :\n",
"\n",
"$z_{i_f}=\\dfrac{x_{i_f}-\\mu_f}{\\sigma_f}$\n",
"\n",
"with : \n",
"- $x_{i_f}$ the value of the feature $f$, for the indivual $i$.\n",
"- $\\mu_f$ the mean of the values for the feature $f$.\n",
"- $\\sigma_f$ the standard deviation of the values for the feature $f$.\n",
"- $z_{i_f}$ the standardized value that we want.\n",
" \n",
"For an instance of `StandardScaler`, the method `fit` compute the mean and the standard deviation to use and the method `transform()` will apply the above formula on each value."
]
},
{
"cell_type": "code",
"execution_count": 17,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:17.662701Z",
"start_time": "2021-10-17T21:24:17.653918Z"
}
},
"outputs": [],
"source": [
"# Separing X and y\n",
"def separe_Xy(data) :\n",
" data_X = data[ X_cols ]\n",
" try :\n",
" data_y = data[ Y_col ]\n",
" except :\n",
" data_y = None\n",
" return data_X, data_y\n",
"\n",
"data1_X, data1_y = separe_Xy(data1)\n",
"data2_X, data2_y = separe_Xy(data2)\n",
"\n",
"# Prevent copy problems\n",
"data1 = None\n",
"data2 = None"
]
},
{
"cell_type": "code",
"execution_count": 18,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:17.670414Z",
"start_time": "2021-10-17T21:24:17.664472Z"
}
},
"outputs": [],
"source": [
"# Standardize features X\n",
"def standardize(data_X, standard_scaler, fit) :\n",
" if fit : data_X = standard_scaler.fit_transform(data_X) # standard_scaler.fit(...) ; standard_scaler.tranfsorm(...)\n",
" else : data_X = standard_scaler.transform(data_X)\n",
" data_X = pd.DataFrame(data_X)\n",
" data_X.columns = X_cols\n",
" return data_X\n",
"\n",
"def destandardize(data_X, standard_scaler) :\n",
" return data_X * np.sqrt(standard_scaler.var_) + standard_scaler.mean_"
]
},
{
"cell_type": "code",
"execution_count": 19,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:17.688680Z",
"start_time": "2021-10-17T21:24:17.672349Z"
},
"scrolled": false
},
"outputs": [],
"source": [
"standard_scaler1 = StandardScaler()\n",
"data1_X = standardize(data1_X,standard_scaler1,fit=True)\n",
"\n",
"standard_scaler2 = StandardScaler()\n",
"data2_X = standardize(data2_X,standard_scaler2,fit=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Preprocessing summary"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Now that we have pre-processed our training data, it is ready to be used to train our model. We now define a function `preprocess` that performs all these preprocessing steps. The reasons is as follows.\n",
"\n",
"After we have train our model on the training dataset, we want to predict labels for a supposed unknown similar dataset. \n",
"\n",
"This supposed unknown dataset is not already preprocessed. We have to: \n",
"\n",
" 1) Rename our columns (only for our convenience)\n",
" 2) Clean our data (removing missing values)\n",
" 3) Standardize our data\n",
"\n",
"\n",
"
\n",
" \n",
"⚠️ Note that standardization use parameters estimated from our training dataset (mean, standard deviation). The unknown similar dataset is supposed to have the same distribution. We don't have to re-estimated these parameters (especially since the training data set is representative). In other terms, we dont't have to apply `fit` method of the `StandardScaler` but only the `transform` method\n",
" \n",
"
\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 20,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:17.699359Z",
"start_time": "2021-10-17T21:24:17.690279Z"
}
},
"outputs": [],
"source": [
"def preprocess_no_scale(data, verbose=False) :\n",
" data = rename_cols(data)\n",
" if verbose : \n",
" print(\"Rename\")\n",
" display(data)\n",
" \n",
" data = remove_empty_lines(data)\n",
" if verbose : \n",
" print(\"Remove empty\")\n",
" display(data)\n",
" \n",
" data = remove_missing(data) \n",
" if verbose : \n",
" print(\"Remove missing\")\n",
" display(data)\n",
" \n",
" data_X, data_y = separe_Xy(data)\n",
" if verbose : \n",
" print(\"Separe X Y\")\n",
" display(data_X)\n",
" display(data_y)\n",
" \n",
" return data_X, data_y\n",
" \n",
"def preprocess_scale(data_X, data_y, standard_scaler, fit, verbose=False) :\n",
" \n",
" data_X = standardize(data_X,standard_scaler,fit)\n",
" if verbose : \n",
" print(\"Standardize X\")\n",
" display(data_X)\n",
" display(data_y)\n",
" \n",
" return data_X, data_y\n",
"\n",
"def preprocess(data, standard_scaler, fit, verbose=False) :\n",
" data_X, data_y = preprocess_no_scale(data, verbose)\n",
" return preprocess_scale(data_X, data_y, standard_scaler, fit, verbose)\n",
" "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# A first model\n",
"\n",
"We will create our models: Random Forest regressors (one model for each turbine). To do this we will instantiate our models, then we will train it with our training data via the `fit` method\n",
"\n",
"## Models creation and training"
]
},
{
"cell_type": "code",
"execution_count": 21,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:21.349847Z",
"start_time": "2021-10-17T21:24:17.702755Z"
}
},
"outputs": [],
"source": [
"def create_and_fit_model(data_X, data_y, model_name, **hyperparameters) :\n",
" regressor = model_name(**hyperparameters)\n",
" regressor.fit(data_X, data_y.values.ravel())\n",
" return regressor\n",
"\n",
"regressor1 = create_and_fit_model(data1_X, data1_y , RandomForestRegressor)\n",
"regressor2 = create_and_fit_model(data2_X, data2_y , RandomForestRegressor)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Testing our models\n",
"\n",
"Now it's time to test our models. For this we will use the test data that we have ignored so far. We do not know this data. The only thing we assume is that the test data set for turbine 1 (respectively, for turbine 2) follows the same probability law as the training data for turbine 1 (respectively, for turbine 2). These are the laws that we have tried to estimate by our models. This is what will allow us to make predictions on the unknown test data.\n",
"\n",
"### Pre-processing of test data\n",
"\n",
"Our test data has not been pre-processed. However, this data must be in the same format as the training data that fed our model. In other words, our test data must not contain any missing values, and the feature values must be standardized using the training parameters (the means and the standard deviations of the training data, assumed to be representative of any data from the wind turbine in question)"
]
},
{
"cell_type": "code",
"execution_count": 22,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:21.379811Z",
"start_time": "2021-10-17T21:24:21.352161Z"
}
},
"outputs": [],
"source": [
"# Preprocess our test data (no fit !)\n",
"data1_X_test, data1_y_test = preprocess(data1_test, standard_scaler1, fit=False)\n",
"data2_X_test, data2_y_test = preprocess(data2_test, standard_scaler2, fit=False)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Making predictions\n",
"\n",
"We can now make our predictions. "
]
},
{
"cell_type": "code",
"execution_count": 23,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:21.541655Z",
"start_time": "2021-10-17T21:24:21.388446Z"
}
},
"outputs": [],
"source": [
"# Make predictions\n",
"data1_y_test_pred = regressor1.predict(data1_X_test)\n",
"data2_y_test_pred = regressor2.predict(data2_X_test)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Performance evaluation.\n",
"\n",
"\n",
"### Defining the metrics\n",
"\n",
"In order to evaluate the success of our predictions, we need to define performance metrics. Since we are dealing with a regression problem, our metrics will measure a distance between the prediction and the true value of the label. For $n$ individuals, with $y_i$ a true label of the individual $i$, $\\hat{y}_i$ the predicted label for $i$ and $\\bar{y}$ the mean of values, we will choose the following metrics: \n",
"\n",
"- R2 Score : score between -1 and 1 (the best possible score is 1)\n",
"\n",
"$R_2 = 1 - \\dfrac{\\sum^{n}_{i=1}{(y_i - \\hat{y}_i)^2}}{\\sum^{n}_{i=1}{(y_i - \\bar{y})^2}}$ \n",
"- Mean Absolute Error : error in the same unit as the label (the best possible score is 0)\n",
"\n",
"$MAE = \\dfrac{\\sum^{n}_{i=1}{|y_i - \\hat{y}_i|}}{n} $\n",
"- Root Mean Squared Error : error in the same unit as the label (the best possible score is 0) \n",
"\n",
"$RMSE = \\displaystyle\\sqrt{\\dfrac{\\sum^{n}_{i=1}{(y_i - \\hat{y}_i)^2}}{n}}$ \n",
"\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Compute scores\n",
"\n",
"We will now calculate the scores of our predictions for each of the models."
]
},
{
"cell_type": "code",
"execution_count": 24,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:21.573742Z",
"start_time": "2021-10-17T21:24:21.545053Z"
}
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Dataset 1 ~ Scores\n",
"R2 Score [-1,1] : \t0.958806564723879\n",
"Mean Absolute Error (kW) : \t40.729163940482145\n",
"Root Mean Squared Error (kW) : \t98.46943075204126\n",
"\n",
"\n",
"\n",
"\n",
"Dataset 2 ~ Scores\n",
"R2 Score [-1,1] : \t0.9861313673199962\n",
"Mean Absolute Error (kW) : \t26.844404390773516\n",
"Root Mean Squared Error (kW) : \t52.913120698991825\n",
"\n"
]
}
],
"source": [
"scores_keys = ['R2 Score [-1,1] ','Mean Absolute Error (kW) ', 'Root Mean Squared Error (kW) ']\n",
"\n",
"def compute_scores(y_pred,y_true) :\n",
" return {\n",
" scores_keys[0] : r2_score(y_pred,y_true),\n",
" scores_keys[1] : mean_absolute_error(y_pred,y_true),\n",
" scores_keys[2] : mean_squared_error(y_pred,y_true,squared=False)\n",
" }\n",
"\n",
"def scores_tostr(y_pred,y_true) :\n",
" s = ''\n",
" scores = compute_scores(y_pred,y_true)\n",
" for idx,score in scores.items() :\n",
" s = s+str(idx)+': \\t'+str(score)+'\\n'\n",
" return s\n",
"\n",
"sco = lambda x : scores_tostr(x[0],x[1])\n",
"print_two(sco,'Scores', (data1_y_test,data1_y_test_pred) , (data2_y_test,data2_y_test_pred))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Plot predictions\n",
"\n",
"Finally, we will plot the actual labels of the test data and the labels predicted by our model as a function of the \"speed\" feature. We choose speed because it is the feature most correlated to the label value."
]
},
{
"cell_type": "code",
"execution_count": 25,
"metadata": {
"ExecuteTime": {
"end_time": "2021-10-17T21:24:22.263111Z",
"start_time": "2021-10-17T21:24:21.578000Z"
}
},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"
"
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"def plot_predictions(scores1, data1_X_test,data1_y_test,data1_y_test_pred,standard_scaler1, model_name1,\n",
" scores2, data2_X_test,data2_y_test,data2_y_test_pred,standard_scaler2, model_name2, \n",
" feature=\"Speed\") :\n",
" \n",
" fig, (ax1,ax2) = plt.subplots(1,2, figsize=(14,8))\n",
" \n",
" if (standard_scaler1 != None) :\n",
" df1 = destandardize(data1_X_test,standard_scaler1)\n",
" else :\n",
" df1 = data1_X_test\n",
" if (standard_scaler2 != None) :\n",
" df2 = destandardize(data2_X_test,standard_scaler2)\n",
" else : \n",
" df2 = data2_X_test\n",
" s = 5\n",
"\n",
" #for col in df1.columns :\n",
"\n",
" ax1.set_title(f\"Dataset 1\")\n",
" ax1.scatter(df1[feature],data1_y_test, c=\"blue\", alpha=0.7, label=\"Real values\", s=s)\n",
" ax1.scatter(df1[feature],data1_y_test_pred, c=\"green\", alpha=0.7, label=\"Predicted values\", s=s)\n",
" ax1.set_ylabel(\"Power\")\n",
" ax1.set_xlabel(\"Speed\")\n",
"\n",
" ax1.text(13,200,f\"R2 Score : { round(scores1[scores_keys[0]],3) }\")\n",
" ax1.text(13,100,f\"MAE : { round(scores1[scores_keys[1]],3) } kW\")\n",
" ax1.text(13,0,f\"RMSE : { round(scores1[scores_keys[2]],3) } kW\")\n",
"\n",
" ax2.set_title(f\"Dataset 2\")\n",
" ax2.scatter(df2[feature],data2_y_test, c=\"red\",alpha=0.7, s=s, label=\"Real values\")\n",
" ax2.scatter(df2[feature],data2_y_test_pred, c=\"green\", alpha=0.7, s=s)\n",
" ax2.set_xlabel(\"Speed\")\n",
" ax2.set_ylabel(\"Power\")\n",
"\n",
" ax2.text(13,200,f\"R2 Score : { round(scores2[scores_keys[0]],3) }\")\n",
" ax2.text(13,100,f\"MAE : { round(scores2[scores_keys[1]],3) } kW\")\n",
" ax2.text(13,0,f\"RMSE : { round(scores2[scores_keys[2]],3) } kW\")\n",
"\n",
" fig.suptitle(f\"{model_name} predictions\")\n",
" fig.legend()\n",
"\n",
" fig.show()\n",
"\n",
"\n",
"model_name = 'Random Forest'\n",
"plot_predictions(compute_scores(data1_y_test_pred,data1_y_test), data1_X_test,data1_y_test,data1_y_test_pred,standard_scaler1, model_name,\n",
" compute_scores(data2_y_test_pred,data2_y_test), data2_X_test,data2_y_test,data2_y_test_pred,standard_scaler2, model_name ) "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### The K-Folds technique\n",
"\n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Measured performance may be biased. Indeed, the separation of the training data and the test data was done in a random way. Maybe the test data and the training data were by chance well chosen in order to obtain such a score. To correct this, we will use the K-folds technique. \n",
"\n",
"First, the data set is randomly separated into k folds. The model will be tested on 1 folds, while the remaining k-1 folds will serve as training data. We reiterate by changing each time the fold that will be used as test data and by noting each time the scores obtained.\n",
"\n",
"
\n",
"\n",
"⚠️ **Important remark:** we do not preprocess the data before separating them into $k$ folds. Indeed, if we preprocess our data before the separation, the standardization would estimate parameters $\\mu$ and $\\sigma$ from the data set. Once this is done, we would separate the data into $k$ folds. One of the folds would serve as a test fold. However, the data of the test fold (supposedly unknown) were taken into account in the calculation of $\\mu$ and $\\sigma$. There would then be a data leak!\n",
" \n",
"Thus, in our `KFolds_validation` function we use non-preprocessed data that we separate into $k$ folds. Within each iteration, we preprocess the training folds: they are standardized. There is no information leakage because the standardization is done on the training folds. After that, we preprocess the test fold by standardizing from the parameters of the training folds (in order to have the test data in the same format), then we make our predictions.\n",
"\n",
"