diff --git a/docs/docs/telco.md b/docs/docs/telco.md
new file mode 100644
index 00000000..f42381de
--- /dev/null
+++ b/docs/docs/telco.md
@@ -0,0 +1,85 @@
+---
+sidebar_position: 4
+---
+
+# Let's build a model predicting company churn
+
+Here we are going to walk through an example where we use scikit.js to predict churn at a major telecomunications company. We will be using the dataset here (Kaggle link).
+
+It's hosted remotely here (though we should probably host it as well).
+
+## Examining the data
+
+First let's create a new directory called `telco` that we are going to perform JS magic in. Let's also initialize npm, so that we can download our packages
+
+> > > mkdir telco
+> > > cd telco
+> > > npm init
+
+For this analysis, we are going to use `danfojs-node`, and `scikitjs`. So let's install them.
+
+> > > npm install danfojs-node scikitjs
+
+```javascript
+import * as dfd from 'danfojs-node'
+import {
+ LogisticRegression,
+ OneHotEncoder,
+ makeColumnTransformer,
+ StandardScaler,
+ makePipeline
+} from 'scikitjs/node'
+
+/**
+ * First let's read in the data and take a look at it.
+ */
+let df = await dfd.readCSV(
+ 'https://raw.githubusercontent.com/rajeevratan84/datascienceforbusiness/master/WA_Fn-UseC_-Telco-Customer-Churn.csv'
+)
+
+/**
+ * The first thing I like to do is figure out the types of the input
+ * We can scour the df.ctypes.print() to see what the types are, and also use
+ * df.
+ * It looks like there are some useless categories.
+ * customerID has only unique values (none are repeated) and so therefor it
+ * can be safely ignored.
+ *
+ * Let's check for any missing values. It looks like there are none.
+ *
+ * By using the unique function we can easily see which columns are categorical
+ * or not. Or my inspecting the danfo column types. df.ctypes.print()
+ */
+
+let categorical = [
+ 'gender',
+ 'Partner',
+ 'Dependents',
+ 'PhoneService',
+ 'Contract',
+ 'PaperlessBilling',
+ 'PaymentMethod',
+ 'MultipleLines',
+ 'InternetService',
+ 'OnlineSecurity',
+ 'TechSupport',
+ 'StreamingTV',
+ 'StreamingMovies',
+ 'OnlineBackup',
+ 'DeviceProtection'
+]
+
+let numerical = ['SeniorCitizen', 'tenure', 'MonthlyCharges', 'TotalCharges']
+
+let y = df['Churn']
+
+let colTransform = makeColumnTransformer([
+ [new OneHotEncoder(), categorical],
+ [new StandardScaler(), numerical]
+])
+
+let lr = makePipeline(colTransform, new LogisticRegression())
+await lr.fit(df, y)
+console.log(lr.coef)
+console.log(lr.score())
+```
pFad - Phonifier reborn
Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies:
Alternative Proxy
pFad Proxy
pFad v3 Proxy
pFad v4 Proxy