add classify-cart nb with intro

Author: jreiberkyle

jupyter-notebooks/crop-classification/classify-cart.ipynb

@@ -0,0 +1,72 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Crop Type Classification: CART\n",
+    "\n",
+    "Our aim in this notebook is to classify crop type using PlanetScope 4-band Orthotiles. The crop types of particular interest are corn and soybeans.\n",
+    "\n",
+    "[CART](http://scikit-learn.org/stable/modules/tree.html#tree-algorithms-id3-c4-5-c5-0-and-cart) is a decision tree algorithm that has shown great promise for classification of remotely sensed imagery. We will use this algorithm to classify crop type.\n",
+    "\n",
+    "In this notebook, we will focus on using the PlanetScope imagery 4 bands as well as NDVI calculation as the features that are fed into the CART algorithm. We will train on one PS Orthotile and validate on another PS Orthotile.\n",
+    "\n",
+    "### Outline\n",
+    "1. Identify datasets (PS Orthotiles and gold standard dataset for train and test)\n",
+    "1. Train classifier\n",
+    "1. Test classifier\n",
+    "\n",
+    "\n",
+    "### Gold Standard Dataset\n",
+    "\n",
+    "The [USDA 2016 Crop Data Layer](https://www.nass.usda.gov/Research_and_Science/Cropland/Release/index.php) (CDL) provides a national crop type dataset. This dataset was build using Landsat 8, DMC Deimos-1, and UK 2 satellite imagery ([src](https://www.nass.usda.gov/Research_and_Science/Cropland/sarsfaqs2.php#Section3_17.0)), using supervised classification (decision trees) based on ground truth from the Farm Service Agency (FSA) Common Land Unit (CLU) Program. Since this is a derived dataset, it isn't a ground truth dataset but it is known to be quite accurate so can be used as our gold standard dataset. This dataset is provided as a georegistered raster (geoTIFF).\n",
+    "\n",
+    "Since corn and soybeans are the primary crops grown in Iowa, we will focus our analysis in that state. The [metadata](https://www.nass.usda.gov/Research_and_Science/Cropland/metadata/metadata_ia16.htm) provided for the Iowa 2016 CDL indicates that it's accuracy is 96.4% and that corn (categorization code 1) and soybeans (categorization code 5) are indeed the primary crop types in the state.\n",
+    "\n",
+    "## Identify Datasets\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 2016 Iowa CDL\n",
+    "\n",
+    "Download data for Iowa from [USDA 2016 Crop Data Layer](https://www.nass.usda.gov/Research_and_Science/Cropland/Release/index.php) (CDL). Filter to corn / soybeans. \n",
+    "\n",
+    "This is achieved by using the [CropScape](https://nassgeodata.gmu.edu/CropScape/) site."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 2",
+   "language": "python",
+   "name": "python2"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 2
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython2",
+   "version": "2.7.13"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}