"Eight more regression models were trained on the same training dataset, namely k nearest neighbors (kNN), linear regression, lasso regression, ridge regression, linear support vector machines (SVM), decision tree, random forest, and gradient boosting machines (GBM).\n",

"\n",

"Using the optimal parameter, the target values were predicted using the test set. The accuracy was measured as the $r^2$ between the true values and predicted values. The table below shows the optimal parameter and corresponding accuracy of each model.\n",

"\n"

"Eight more regression models were trained on the same training dataset, namely k nearest neighbors (kNN), linear regression, lasso regression, ridge regression, linear support vector machines (LSVM), decision tree, random forest, and gradient boosting machines (GBM).\n",

"\n",

"Using the optimal parameter, the target values were predicted using the test set. The accuracy was measured as the $r^2$ between the true values and predicted values. The table below shows the optimal parameter of each model.\n",

"\n",

"|Model|Parameters|\n",

"|:-|:-|\n",

"|Feed-forward NN|no. of nodes<br>no. of hidden layers<br>activation functions<br>learning rates\n",

"|kNN|no. of nearest neighbors|\n",

"|Linear regression|-|\n",

"|Lasso regression|alpha|\n",

"|Ridge regression|alpha|\n",

"|LSVM|C|\n",

"|Decision tree|max depth|\n",

"|Random forest|max depth|\n",

"|GBM|max depth|"

]

},

{

...

...

@@ -102,9 +113,21 @@

"cell_type": "markdown",

"metadata": {},

"source": [

"The feed forward neural network was able to predict bike-sharing counts with .... accuracy, higher than the eight other machine learning models trained. The table below summarizes the predictive accuracies and corresponding parameters of all models evaluated:\n",

"The feed forward neural network was able to predict bike-sharing counts on the test set with 74.2% accuracy, higher than the eight other machine learning models trained. The table below summarizes the predictive accuracies and corresponding parameters of all models evaluated\n",

"\n",

"|Model|Parameters|Values|Test Accuracy|\n",

"|:-|:-|:-:|:-|\n",

"|Feed-forward NN|no. of nodes<br>no. of hidden layers<br>activation functions<br>learning rates|(56, 56, 1) <br>1 <br>(linear, sine, sigmoid) <br>(0.001, 0.0001)|74.2%|\n",

"|GBM|max depth|19|73.4%|\n",

"|Random forest|max depth|31|72.7%|\n",

"|kNN|no. of nearest neighbors|4|68.5%\n",

"|Decision tree|max depth|27|50.5%\n",

"|Lasso regression|alpha|0.0001|44.8%\n",

"|Ridge regression|alpha|10|41.9%\n",

"|Linear regression|-|-|41.3%\n",

"|LSVM|C|0.1|33.0%\n",

"\n",

"|TABLE|OPTIMAL PARAMS AND ACCURACIES|"

"\n"

]

},

{

...

...

@@ -114,6 +137,13 @@

"## Conclusion"

]

},

{

"cell_type": "markdown",

"metadata": {},

"source": [

"Bike sharing data were taken from XXXX. Hourly counts of bike shares were predicted most accurately using a feed-forward neural network, followed by GBM and random forest."