Updated file paths

01_Make_models.ipynb

@@ -2,7 +2,7 @@
  "cells": [
   {
    "cell_type": "markdown",
-   "id": "03d50317",
+   "id": "successful-responsibility",
    "metadata": {},
    "source": [
     "## Code accompanyment to\"Machine learning applied to a modern-Pleistocene petrographic dataset: The global prediction of sand modal composition (GloPrSM) model\"\n",
@@ -13,15 +13,15 @@
   },
   {
    "cell_type": "markdown",
-   "id": "61b900a0",
+   "id": "stone-headset",
    "metadata": {},
    "source": [
     "## Step 1: Load sand modal composition data and make random forests models"
    ]
   },
   {
    "cell_type": "markdown",
-   "id": "8fd11344",
+   "id": "differential-exposure",
    "metadata": {},
    "source": [
     "### Import required modules"
@@ -30,7 +30,7 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "1163c19f",
+   "id": "tender-heather",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -44,12 +44,13 @@
     "from sklearn.metrics import *\n",
     "import time\n",
     "import pickle\n",
-    "import pathlib"
+    "import pathlib\n",
+    "import os"
    ]
   },
   {
    "cell_type": "markdown",
-   "id": "ca3bdcdc",
+   "id": "hollow-contribution",
    "metadata": {},
    "source": [
     "### Load dependent variable data"
@@ -58,7 +59,7 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "86b7b6be",
+   "id": "inclusive-moral",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -68,7 +69,7 @@
   },
   {
    "cell_type": "markdown",
-   "id": "9fde66b0",
+   "id": "prescription-furniture",
    "metadata": {},
    "source": [
     "### Feature selection"
@@ -77,7 +78,7 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "bf001f1e",
+   "id": "pointed-blink",
    "metadata": {},
    "outputs": [],
    "source": [
@@ -89,7 +90,7 @@
   },
   {
    "cell_type": "markdown",
-   "id": "4219c186",
+   "id": "suffering-black",
    "metadata": {},
    "source": [
     "### Feature correlation (optional)"
@@ -98,7 +99,7 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "408ed37d",
+   "id": "voluntary-accuracy",
    "metadata": {
     "scrolled": true
    },
@@ -114,7 +115,7 @@
   },
   {
    "cell_type": "markdown",
-   "id": "d2edafd3",
+   "id": "selective-major",
    "metadata": {},
    "source": [
     "### Model training\n",
@@ -124,16 +125,16 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "bed73d58",
+   "id": "spectacular-formation",
    "metadata": {},
    "outputs": [],
    "source": [
-    "# Specify output folder\n",
-    "base_path = r'Z:\\Sharman\\GloPrSM_git\\v1.0'\n",
+    "# Specify output folder (current directory used by default)\n",
+    "base_path = os.getcwd() + '/v1.0'\n",
     "\n",
-    "model_path = base_path + '\\\\' + 'models'\n",
-    "val_path = base_path + '\\\\' + 'validation'\n",
-    "test_path = base_path + '\\\\' + 'test_labels'\n",
+    "model_path = base_path + '/' + 'models'\n",
+    "val_path = base_path + '/' + 'validation'\n",
+    "test_path = base_path + '/' + 'test_labels'\n",
     "\n",
     "# Recursively creates the directory and does not raise an exception if the directory already exists\n",
     "pathlib.Path(model_path).mkdir(parents=True, exist_ok=True)\n",
@@ -144,13 +145,13 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "e727d4dd",
+   "id": "technological-villa",
    "metadata": {},
    "outputs": [],
    "source": [
     "labels = ['FQ_QFL_IJ', 'LQ_QFL_IJ', 'QmQch_QmQpQch_IJ', 'QpQch_QmQpQch_IJ', 'FkFp_FpFk_IJ', 'LsLv_LvLsLm_IJ', 'LmLv_LvLsLm_IJ']\n",
     "\n",
-    "splits = 10 # Note, 100 splits are used in the article\n",
+    "splits = 1 # Note, 100 splits are used in the article\n",
     "rs = ShuffleSplit(n_splits=splits, test_size=.2, random_state=0)\n",
     "stats = np.zeros(shape=(splits,len(labels)))\n",
     "\n",
@@ -184,27 +185,35 @@
     "        tst_df.loc[:,'{}_label_{}'.format(label, i)] = test_labels\n",
     "\n",
     "        # Export the validation results\n",
-    "        val_df.to_csv(val_path+'\\\\'+'{}_validation_rlf.csv'.format(label),index=False)\n",
-    "        tst_df.to_csv(test_path+'\\\\'+'{}_label_rlf.csv'.format(label),index=False)\n",
+    "        val_df.to_csv(val_path+'/'+'{}_validation_rlf.csv'.format(label),index=False)\n",
+    "        tst_df.to_csv(test_path+'/'+'{}_label_rlf.csv'.format(label),index=False)\n",
     "        \n",
     "        # Save the model\n",
     "        model_filename = 'model_'+str(i)+'.sav'\n",
-    "        model_filepath = model_path+'\\\\'+str(label)\n",
+    "        model_filepath = model_path+'/'+str(label)\n",
     "        pathlib.Path(model_filepath).mkdir(parents=True, exist_ok=True) # Recursively creates the directory and does not raise an exception if the directory already exists\n",
-    "        pickle.dump(model, open(model_filepath+'\\\\'+model_filename, 'wb'))\n",
+    "        pickle.dump(model, open(model_filepath+'/'+model_filename, 'wb'))\n",
     "            \n",
     "        print(i, 'R2: {}, {} sec'.format(round(r2,6), round(time.time()-start,1)))    \n",
     "        i += 1\n",
     "    print()\n",
     "\n",
     "r2_df = pd.DataFrame(stats, columns=[x+'_R2' for x in labels])\n",
-    "r2_df.to_csv(base_path+'\\\\'+'R2_stats_rlf.csv',index=False)\n",
+    "r2_df.to_csv(base_path+'/'+'R2_stats_rlf.csv',index=False)\n",
     "\n",
     "# Save the list of features used in the models, so you know what is going on\n",
     "features = pd.DataFrame()\n",
     "features['Inputs'] = feature_list\n",
-    "features.to_csv(base_path+'\\\\'+'feature_list.csv', index=False)"
+    "features.to_csv(base_path+'/'+'feature_list.csv', index=False)"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "referenced-friend",
+   "metadata": {},
+   "outputs": [],
+   "source": []
   }
  ],
  "metadata": {