Data_Mining_Final_Project.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "ir",
      "display_name": "R"
    },
    "language_info": {
      "name": "R"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ctEdp9ATFT8A"
      },
      "outputs": [],
      "source": [
        "#install.packages(c(\"readr\", \"rpart\", \"rpart.plot\", \"survival\", \"survminer\" , \"ROCR\" ,\"e1071\"))\n",
        "library(readr)\n",
        "library(rpart)\n",
        "library(rpart.plot)\n",
        "library(survival)\n",
        "library(survminer)\n",
        "library(lattice)\n",
        "library(ggplot2)\n",
        "library(caret)\n",
        "library(ROCR)\n",
        "library(e1071)\n",
        "\n",
        "\n",
        "\n",
        "df <- read.csv(\"telco.csv\")"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Data Cleaning/Manipulation"
      ],
      "metadata": {
        "id": "Nm0i45dBFavF"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Drop the 'customerID' column\n",
        "df <- subset(df, select = -customerID)\n",
        "\n",
        "# Convert 'TotalCharges' to numeric\n",
        "df$TotalCharges <- as.numeric(as.character(df$TotalCharges))\n",
        "\n",
        "# Check for missing values - TotalCharges has 11\n",
        "print(colSums(is.na(df)))\n",
        "\n",
        "# Impute missing values in 'TotalCharges' with mean\n",
        "df$TotalCharges[is.na(df$TotalCharges)] <- mean(df$TotalCharges, na.rm = TRUE)\n",
        "\n",
        "# Drop rows where 'tenure' is 0\n",
        "rows_with_tenure_0 <- which(df$tenure == 0)\n",
        "df <- df[-rows_with_tenure_0, ]\n",
        "\n",
        "# Convert 'SeniorCitizen' to categorical\n",
        "df$SeniorCitizen <- factor(df$SeniorCitizen, levels = c(0, 1), labels = c(\"No\", \"Yes\"))\n",
        "\n",
        "# Create a list of categorical columns\n",
        "cat_cols <- names(df)[sapply(df, is.factor)]\n",
        "cat_cols <- cat_cols[cat_cols != \"customerID\"]  # Remove 'customerID' column\n",
        "\n",
        "# Convert categorical variables to factors\n",
        "df$gender <- as.factor(df$gender)\n",
        "df[cat_cols] <- lapply(df[cat_cols], as.factor)\n",
        "\n",
        "# Display summary statistics for 'InternetService'\n",
        "print(summary(df$InternetService))\n",
        "\n",
        "# Display summary statistics for numerical columns\n",
        "numerical_cols <- c('tenure', 'MonthlyCharges', 'TotalCharges')\n",
        "print(summary(df[, numerical_cols]))\n",
        "\n",
        "# Convert 'Churn' to factor\n",
        "df$Churn <- factor(df$Churn, levels = c(\"No\", \"Yes\"))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "j9VlTNjNFeA4",
        "outputId": "ace347ed-e7c0-4ce2-f47d-4fd2d7c996ed"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "          gender    SeniorCitizen          Partner       Dependents \n",
            "               0                0                0                0 \n",
            "          tenure     PhoneService    MultipleLines  InternetService \n",
            "               0                0                0                0 \n",
            "  OnlineSecurity     OnlineBackup DeviceProtection      TechSupport \n",
            "               0                0                0                0 \n",
            "     StreamingTV  StreamingMovies         Contract PaperlessBilling \n",
            "               0                0                0                0 \n",
            "   PaymentMethod   MonthlyCharges     TotalCharges            Churn \n",
            "               0                0               11                0 \n",
            "   Length     Class      Mode \n",
            "     7032 character character \n",
            "     tenure      MonthlyCharges    TotalCharges   \n",
            " Min.   : 1.00   Min.   : 18.25   Min.   :  18.8  \n",
            " 1st Qu.: 9.00   1st Qu.: 35.59   1st Qu.: 401.4  \n",
            " Median :29.00   Median : 70.35   Median :1397.5  \n",
            " Mean   :32.42   Mean   : 64.80   Mean   :2283.3  \n",
            " 3rd Qu.:55.00   3rd Qu.: 89.86   3rd Qu.:3794.7  \n",
            " Max.   :72.00   Max.   :118.75   Max.   :8684.8  \n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Kaplan-Meier Model on two cohorts of customers based on whether a customer has subscribed for Streaming TV or not."
      ],
      "metadata": {
        "id": "vodfvOXg0egk"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "kmf <- survfit(Surv(tenure, Churn) ~ StreamingTV, data = df)\n",
        "kmf"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 196
        },
        "id": "W_3Cl3oOz9Xz",
        "outputId": "e5bc7f30-9e40-49ca-ea83-88441f3eab21"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "Call: survfit(formula = Surv(tenure, Churn) ~ StreamingTV, data = df)\n",
              "\n",
              "                                         n nevent    rmean*\n",
              "StreamingTV=No, (s0)                  2809      0 47.440355\n",
              "StreamingTV=No internet service, (s0) 1520      0 66.459816\n",
              "StreamingTV=Yes, (s0)                 2703      0 56.068951\n",
              "StreamingTV=No, Yes                   2809    942 24.559645\n",
              "StreamingTV=No internet service, Yes  1520    113  5.540184\n",
              "StreamingTV=Yes, Yes                  2703    814 15.931049\n",
              "   *restricted mean time in state (max time = 72 )"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "- n: number of observations in each group.\n",
        "- nevent: churn observed in each group.\n",
        "- rmean*: restricted mean time in state, which is the average time until churn occurs or the study ends.\n",
        "\n",
        "Customers who are not subscribed to streaming TV or have no internet service have longer average times until churn compared to those who are subscribed to streaming TV or have internet service.\n",
        "\n",
        "Customers who are subscribed to both streaming TV and internet service tend to have shorter average times until churn compared to other groups.\n"
      ],
      "metadata": {
        "id": "MDTumVa5z_aR"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Train - Test Split"
      ],
      "metadata": {
        "id": "bICR-EcQ5N3Y"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Train - Test\n",
        "set.seed(123)\n",
        "\n",
        "# Split data, 75% distribution of churn for training\n",
        "train.index <- createDataPartition( y = df$Churn, p = 0.75, list = FALSE)\n",
        "train <- df[train.index,]\n",
        "test <- df[-train.index,]"
      ],
      "metadata": {
        "id": "SFTWPn2N5RbH"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Classification Tree Model"
      ],
      "metadata": {
        "id": "TZUQH7rBFlXg"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "tree_model <- rpart(Churn ~ ., data = df, method = \"class\")\n",
        "\n",
        "# Visualize the tree\n",
        "rpart.plot(tree_model)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 437
        },
        "id": "36WR1Q9WFqCu",
        "outputId": "9e9d0440-c4ef-46a0-c972-8c79d0f0b79f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "plot without title"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94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\nAElFTkSuQmCC"
          },
          "metadata": {
            "image/png": {
              "width": 420,
              "height": 420
            }
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Classification Tree - Confusion Matrix - Accuracy\n",
        "\n",
        "# Prediction, classification  -  Classificatin Tree\n",
        "tree.pred <- predict(tree_model,test, type = \"class\")\n",
        "confusionMatrix(tree.pred,test$Churn)\n",
        "\n",
        "# Prediction, probability\n",
        "tree.pred.prob <- predict(tree_model, test, type = \"prob\")\n",
        "tree.pred.prob.val <- prediction(tree.pred.prob[,2],test$Churn)\n",
        "tree.pred.prob.perf <- performance(tree.pred.prob.val, \"auc\")\n",
        "\n",
        "# print AUC value\n",
        "paste(\"AUC Value is:\", as.numeric(performance(tree.pred.prob.val, \"auc\")@y.values))\n",
        "# plots the ROC curve with colors where the splits are.\n",
        "plot(performance(tree.pred.prob.val, \"tpr\", \"fpr\"), colorize = TRUE ,main = \"ROC Curve - Classification Tree\")"
      ],
      "metadata": {
        "id": "UHpfTBJ15IwB"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Random Forest"
      ],
      "metadata": {
        "id": "QUFpThId5bKC"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#Random Forest\n",
        "\n",
        "rf.fit  <- train(Churn ~ . , method = \"rf\", data = train, importance = TRUE)\n",
        "rf.pred <- predict(rf.fit, test)\n",
        "confusionMatrix(rf.pred, test$Churn)\n",
        "#performance(rf.pred, measure = \"auc\")@y.values[[1]]\n",
        "\n",
        "# Prediction, probability\n",
        "rf.pred.prob <- predict(rf.fit, test, type = \"prob\")\n",
        "rf.pred.prob.val <- prediction(rf.pred.prob[,2],test$Churn)\n",
        "rf.pred.prob.perf <- performance(rf.pred.prob.val,\"auc\")\n",
        "\n",
        "# print AUC value - Random Forest Model\n",
        "paste(\"AUC Value is:\", as.numeric(performance(rf.pred.prob.val, \"auc\")@y.values))\n",
        "# plots the ROC curve with colors where the splits are.\n",
        "plot(performance(rf.pred.prob.val, \"tpr\", \"fpr\"), colorize = TRUE , main = \"ROC Curve - Random Forest\")"
      ],
      "metadata": {
        "id": "J8xZTTua5ece"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# SVM"
      ],
      "metadata": {
        "id": "hHxJbKKW5kNG"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model_svm <- svm(Churn~., train, kernel=\"linear\", cost=0.1)\n",
        "model_svm\n",
        "predicted_svm<-predict(model_svm,test,decision.values = TRUE)\n",
        "mean(predicted_svm==test$Churn)\n",
        "\n",
        "svm_ct <- table(predicted_svm, test$Churn)\n",
        "svm_ct\n",
        "svm_recall <- svm_ct[2,2]/(svm_ct[2,2]+svm_ct[1,2])\n",
        "svm_recall"
      ],
      "metadata": {
        "id": "8KnFZ_VJ5aRX"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Linear Regression"
      ],
      "metadata": {
        "id": "jUtp4GKlrWUe"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "full.model <- glm(Churn ~., train, family = binomial)\n",
        "summary(full.model)\n",
        "\n",
        "\n",
        "step.model <- step(full.model, direction = \"both\", trace = FALSE)\n",
        "summary(step.model)\n",
        "\n",
        "#Confusion Matrix for step.model\n",
        "\n",
        "predict.test <- predict(step.model, newdata = test, type = 'response')\n",
        "predict.train <- predict(step.model, newdata = train, type = 'response')\n",
        "\n",
        "# Convert predicted probabilities to class labels\n",
        "predicted_classes_test <- ifelse(predict.test > 0.5, \"Yes\", \"No\")\n",
        "predicted_classes_train <- ifelse(predict.train > 0.5, \"Yes\", \"No\")\n",
        "\n",
        "# Get the actual class labels\n",
        "true_labels_test <- test$Churn\n",
        "true_labels_train <- train$Churn\n",
        "\n",
        "# Compute the confusion matrix\n",
        "\n",
        "confusion_matrix_test <- table(true_labels, predicted_classes_test)\n",
        "print(paste(\"Confusion Matrix Test\", confusion_matrix_test))\n",
        "\n",
        "confusion_matrix_train <- table(true_labels_train, predicted_classes_train)\n",
        "print(paste(\"Confusion Matrix Train\", confusion_matrix_train))\n",
        "\n",
        "# Accuracy of the model:\n",
        "accuracy_test <- sum(predicted_classes_test == true_labels_test) / length(true_labels_test)\n",
        "accuracy_train <- sum(predicted_classes_train == true_labels_train) / length(true_labels_train)"
      ],
      "metadata": {
        "id": "hJTt8bKyrbE8"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## KNN"
      ],
      "metadata": {
        "id": "Cyqdx4wn2Bx6"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "preProcValues <- preProcess(train, method = c(\"center\", \"scale\"))\n",
        "trainTransformed <- predict(preProcValues, train)\n",
        "testTransformed <- predict(preProcValues, test)"
      ],
      "metadata": {
        "id": "PCh9M_aIixld"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "## K = 3\n",
        "knn.churn3<- knn3(\n",
        "  Churn ~ gender + SeniorCitizen + Partner + Dependents + tenure + PhoneService + MonthlyCharges + TotalCharges,\n",
        "  data = trainTransformed,\n",
        "  k = 3\n",
        ")\n",
        "\n",
        "predictions3 <- predict(knn.churn3, testTransformed,type = \"class\")\n",
        "\n",
        "# Calculate confusion matrix\n",
        "cm3 <- confusionMatrix(predictions3, testTransformed$Churn)\n",
        "cm3\n",
        "\n",
        "results3 <- data.frame(Accuracy = cm3$overall[\"Accuracy\"],\n",
        "           Sensitivity = cm3$byClass[\"Sensitivity\"],\n",
        "           Specificity = cm3$byClass[\"Specificity\"])\n",
        "results3$Accuracy\n"
      ],
      "metadata": {
        "id": "wTWv45Fu2L0v"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "knn.churn5<- knn3(\n",
        "  Churn ~ gender + SeniorCitizen + Partner + Dependents + tenure + PhoneService + MonthlyCharges + TotalCharges,\n",
        "  data = trainTransformed,\n",
        "  k = 5\n",
        ")\n",
        "\n",
        "predictions5 <- predict(knn.churn5, testTransformed,type = \"class\")\n",
        "\n",
        "# Calculate confusion matrix\n",
        "cm5 <- confusionMatrix(predictions5, testTransformed$Churn)\n",
        "cm5\n",
        "\n",
        "results5 <- data.frame(Accuracy = cm5$overall[\"Accuracy\"],\n",
        "                       Sensitivity = cm5$byClass[\"Sensitivity\"],\n",
        "                       Specificity = cm5$byClass[\"Specificity\"])\n",
        "results5$Accuracy"
      ],
      "metadata": {
        "id": "JgxdFA1r2OD6"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "##KNN k = 9\n",
        "knn.churn9<- knn3(\n",
        "  Churn ~ gender + SeniorCitizen + Partner + Dependents + tenure + PhoneService + MonthlyCharges + TotalCharges,\n",
        "  data = trainTransformed,\n",
        "  k = 9\n",
        ")\n",
        "\n",
        "predictions9 <- predict(knn.churn9, testTransformed,type = \"class\")\n",
        "\n",
        "# Calculate confusion matrix\n",
        "cm9 <- confusionMatrix(predictions9, testTransformed$Churn)\n",
        "cm9\n",
        "\n",
        "results9 <- data.frame(Accuracy = cm9$overall[\"Accuracy\"],\n",
        "                       Sensitivity = cm9$byClass[\"Sensitivity\"],\n",
        "                       Specificity = cm9$byClass[\"Specificity\"])\n",
        "results9$Accuracy"
      ],
      "metadata": {
        "id": "t4Az50Ez2QmM"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "## KNN k = 11\n",
        "knn.churn11<- knn3(\n",
        "  Churn ~ gender + SeniorCitizen + Partner + Dependents + tenure + PhoneService + MonthlyCharges + TotalCharges,\n",
        "  data = trainTransformed,\n",
        "  k = 11\n",
        ")\n",
        "\n",
        "predictions11 <- predict(knn.churn11, testTransformed,type = \"class\")\n",
        "\n",
        "# Calculate confusion matrix\n",
        "cm11 <- confusionMatrix(predictions11, testTransformed$Churn)\n",
        "cm11\n",
        "\n",
        "results11 <- data.frame(Accuracy = cm11$overall[\"Accuracy\"],\n",
        "                       Sensitivity = cm11$byClass[\"Sensitivity\"],\n",
        "                       Specificity = cm11$byClass[\"Specificity\"])\n",
        "results11$Accuracy"
      ],
      "metadata": {
        "id": "Clz4c29D2SjA"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "## KNN k = 13\n",
        "knn.churn13<- knn3(\n",
        "  Churn ~ gender + SeniorCitizen + Partner + Dependents + tenure + PhoneService + MonthlyCharges + TotalCharges,\n",
        "  data = trainTransformed,\n",
        "  k = 13\n",
        ")\n",
        "\n",
        "predictions13 <- predict(knn.churn13,\n",
        "testTransformed,type = \"class\")\n",
        "# Calculate confusion matrix\n",
        "cm13 <- confusionMatrix(predictions13,testTransformed$Churn)\n",
        "cm13\n",
        "\n",
        "results13 <- data.frame(Accuracy = cm13$overall[\"Accuracy\"],\n",
        "                        Sensitivity = cm13$byClass[\"Sensitivity\"],\n",
        "                        Specificity = cm13$byClass[\"Specificity\"])\n",
        "results13$Accuracy"
      ],
      "metadata": {
        "id": "5m_y-Tar2eyi"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "## KNN k = 15\n",
        "knn.churn15<- knn3(\n",
        "  Churn ~ gender + SeniorCitizen + Partner + Dependents + tenure + PhoneService + MonthlyCharges + TotalCharges,\n",
        "  data = trainTransformed,\n",
        "  k = 15\n",
        ")\n",
        "\n",
        "predictions15 <- predict(knn.churn15, testTransformed,type = \"class\")\n",
        "\n",
        "# Calculate confusion matrix\n",
        "cm15 <- confusionMatrix(predictions15, testTransformed$Churn)\n",
        "cm15\n",
        "\n",
        "results15 <- data.frame(Accuracy = cm15$overall[\"Accuracy\"],\n",
        "                        Sensitivity = cm15$byClass[\"Sensitivity\"],\n",
        "                        Specificity = cm15$byClass[\"Specificity\"])\n",
        "results15$Accuracy"
      ],
      "metadata": {
        "id": "tXEMxM2J2hJ9"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "## KNN k = 19\n",
        "knn.churn19<- knn3(\n",
        "  Churn ~ gender + SeniorCitizen + Partner + Dependents + tenure + PhoneService + MonthlyCharges + TotalCharges,\n",
        "  data = trainTransformed,\n",
        "  k = 19\n",
        ")\n",
        "\n",
        "predictions19 <- predict(knn.churn19, testTransformed,type = \"class\")\n",
        "\n",
        "# Calculate confusion matrix\n",
        "cm19 <- confusionMatrix(predictions19, testTransformed$Churn)\n",
        "cm19\n",
        "\n",
        "results19 <- data.frame(Accuracy = cm19$overall[\"Accuracy\"],\n",
        "                        Sensitivity = cm19$byClass[\"Sensitivity\"],\n",
        "                        Specificity = cm19$byClass[\"Specificity\"])\n",
        "results19$Accuracy"
      ],
      "metadata": {
        "id": "NC5eWhQq2lrb"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "## Confirming K = 11 is best value\n",
        "knnModel <- train(\n",
        "  Churn ~ gender + SeniorCitizen + Partner + Dependents + tenure + PhoneService + MonthlyCharges + TotalCharges,\n",
        "  data = trainTransformed,\n",
        "  method = \"knn\",\n",
        "  trControl = trainControl(method = \"cv\"),\n",
        "  tuneGrid = data.frame(k = c(3,5,9,11,13,15,19))\n",
        ")\n",
        "\n",
        "best_model<- knn3(\n",
        "  Churn ~ gender + SeniorCitizen + Partner + Dependents + tenure + PhoneService + MonthlyCharges + TotalCharges,\n",
        "  data = trainTransformed,\n",
        "  k = knnModel$bestTune$k\n",
        ")\n",
        "\n",
        "accuracy.df <- data.frame(\n",
        "  k_value = c(3,5,9,11,13,15,19),\n",
        "  accuracy = c(results3$Accuracy,results5$Accuracy,results9$Accuracy, results11$Accuracy, results13$Accuracy, results15$Accuracy, results19$Accuracy)\n",
        ")\n",
        "accuracy.df\n",
        "\n",
        "p <- ggplot(accuracy.df, aes(x = k_value, y = accuracy)) +\n",
        "  geom_line(color = \"lightgreen\", linewidth = 2) +\n",
        "  geom_point(color = \"darkgreen\", size = 5) +\n",
        "  labs(title = \"K-Nearest Neighbors Accuracy for Different K Values\",\n",
        "       x = \"Number of Neighbors (K Value)\",\n",
        "       y = \"Accuracy\") +\n",
        "  theme_minimal()\n",
        "\n",
        "p + theme(plot.title = element_text(hjust = .5, face = 'bold'))"
      ],
      "metadata": {
        "id": "ui5AYUq62nvU"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}