diff --git a/Dataset/readme.md b/Dataset/readme.md
new file mode 100644
index 00000000..8eb378af
--- /dev/null
+++ b/Dataset/readme.md
@@ -0,0 +1,2 @@
+Dataset used in this project can be accessed from this link:
+https://drive.google.com/file/d/1MYD5xZvPlUdi3WrWjhgOXpWQdAqJ4rlc/view?usp=sharing
diff --git a/TechTitans/readme.md b/TechTitans/readme.md
new file mode 100644
index 00000000..47332373
--- /dev/null
+++ b/TechTitans/readme.md
@@ -0,0 +1,4 @@
+Team TechTitans
+Members: Jyotish Ranjan Mallik, Swarupa Das, Subhashree Dutta, Somesh Kant
+[TechTitans.pptx](https://github.com/SubhashreeDutta/CISHack-UITBU/files/6270072/TechTitans.pptx)
+[TechTitans.txt](https://github.com/SubhashreeDutta/CISHack-UITBU/files/6270074/TechTitans.txt)
diff --git a/TechTitans_(Solution).ipynb b/TechTitans_(Solution).ipynb
new file mode 100644
index 00000000..2042e7e1
--- /dev/null
+++ b/TechTitans_(Solution).ipynb
@@ -0,0 +1,3223 @@
+{
+ "nbformat": 4,
+ "nbformat_minor": 0,
+ "metadata": {
+ "kernelspec": {
+ "display_name": "Python 3",
+ "language": "python",
+ "name": "python3"
+ },
+ "language_info": {
+ "codemirror_mode": {
+ "name": "ipython",
+ "version": 3
+ },
+ "file_extension": ".py",
+ "mimetype": "text/x-python",
+ "name": "python",
+ "nbconvert_exporter": "python",
+ "pygments_lexer": "ipython3",
+ "version": "3.7.3"
+ },
+ "colab": {
+ "name": "TechTitans (Solution).ipynb",
+ "provenance": [],
+ "include_colab_link": true
+ }
+ },
+ "cells": [
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "view-in-github",
+ "colab_type": "text"
+ },
+ "source": [
+ ""
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "V2_8YwM_RML7"
+ },
+ "source": [
+ "# ML Model to filter out False Positive generated by Alert system."
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "WTOXDDBdRMMC"
+ },
+ "source": [
+ "## Context\n",
+ "There is a lack of public available datasets on financial services and specially in the emerging mobile money transactions domain. Financial datasets are important to many researchers and in particular to us performing research in the domain of fraud detection. Part of the problem is the intrinsically private nature of financial transactions, that leads to no publicly available datasets.\n",
+ "\n",
+ "We present a synthetic dataset generated using the simulator called PaySim as an approach to such a problem. PaySim uses aggregated data from the private dataset to generate a synthetic dataset that resembles the normal operation of transactions and injects malicious behaviour to later evaluate the performance of fraud detection methods."
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "-MQZ9h-yRMMD"
+ },
+ "source": [
+ "## Predicting True Positive from Fraudlent Transactions:\n",
+ "\n",
+ "We are presented with a labeled dataset of financial transactions, some of which are case of money laundering. We will be performing exploratory data analysis on this data, and then creating a classifier model to predict whether a transaction is real alert or not, given the included features. The objective of this project is to explain my thought processes in solving this problem, as well as addressing some of the issues that inherently face machine learning models. (\"All models are wrong, but some are useful.\") Using this notebook, I hope to focus primarily on transparency and clarity rather than raw predictive performance, and readability for an audience without a specialization in data science.\n",
+ "\n",
+ "'Note:' Here whenever we talk about fraud or fraudulent, it means alert system marked it as Money laundering. "
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "D7VyevGHRMME"
+ },
+ "source": [
+ "### 1. Data Input and EDA"
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "xHlhpuLQRMME"
+ },
+ "source": [
+ "import numpy as np\n",
+ "import pandas as pd"
+ ],
+ "execution_count": null,
+ "outputs": []
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "mRlkf2-eRMMF"
+ },
+ "source": [
+ "df = pd.read_csv('PS_20174392719_1491204439457_log.csv')"
+ ],
+ "execution_count": null,
+ "outputs": []
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "VPxqTYt3RMMF"
+ },
+ "source": [
+ "The data has about 6.3 million rows and 11 columns. Each row represents a transaction."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "mkIq3bGGRMMF",
+ "outputId": "75c98b38-cf4d-4c4e-f897-5183bc3e1c60"
+ },
+ "source": [
+ "df.shape"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "execute_result",
+ "data": {
+ "text/plain": [
+ "(6362620, 11)"
+ ]
+ },
+ "metadata": {
+ "tags": []
+ },
+ "execution_count": 3
+ }
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "vn3b3Wy9RMMG",
+ "outputId": "72b02c36-d8c0-4c58-c5e2-477a6cf68d0b"
+ },
+ "source": [
+ "df.head()"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "execute_result",
+ "data": {
+ "text/html": [
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ "
step
\n",
+ "
type
\n",
+ "
amount
\n",
+ "
nameOrig
\n",
+ "
oldbalanceOrg
\n",
+ "
newbalanceOrig
\n",
+ "
nameDest
\n",
+ "
oldbalanceDest
\n",
+ "
newbalanceDest
\n",
+ "
isFraud
\n",
+ "
isFlaggedFraud
\n",
+ "
\n",
+ " \n",
+ " \n",
+ "
\n",
+ "
0
\n",
+ "
1
\n",
+ "
PAYMENT
\n",
+ "
9839.64
\n",
+ "
C1231006815
\n",
+ "
170136.0
\n",
+ "
160296.36
\n",
+ "
M1979787155
\n",
+ "
0.0
\n",
+ "
0.0
\n",
+ "
0
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
1
\n",
+ "
1
\n",
+ "
PAYMENT
\n",
+ "
1864.28
\n",
+ "
C1666544295
\n",
+ "
21249.0
\n",
+ "
19384.72
\n",
+ "
M2044282225
\n",
+ "
0.0
\n",
+ "
0.0
\n",
+ "
0
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
2
\n",
+ "
1
\n",
+ "
TRANSFER
\n",
+ "
181.00
\n",
+ "
C1305486145
\n",
+ "
181.0
\n",
+ "
0.00
\n",
+ "
C553264065
\n",
+ "
0.0
\n",
+ "
0.0
\n",
+ "
1
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
3
\n",
+ "
1
\n",
+ "
CASH_OUT
\n",
+ "
181.00
\n",
+ "
C840083671
\n",
+ "
181.0
\n",
+ "
0.00
\n",
+ "
C38997010
\n",
+ "
21182.0
\n",
+ "
0.0
\n",
+ "
1
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
4
\n",
+ "
1
\n",
+ "
PAYMENT
\n",
+ "
11668.14
\n",
+ "
C2048537720
\n",
+ "
41554.0
\n",
+ "
29885.86
\n",
+ "
M1230701703
\n",
+ "
0.0
\n",
+ "
0.0
\n",
+ "
0
\n",
+ "
0
\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
"
+ ],
+ "text/plain": [
+ " step type amount nameOrig oldbalanceOrg newbalanceOrig \\\n",
+ "0 1 PAYMENT 9839.64 C1231006815 170136.0 160296.36 \n",
+ "1 1 PAYMENT 1864.28 C1666544295 21249.0 19384.72 \n",
+ "2 1 TRANSFER 181.00 C1305486145 181.0 0.00 \n",
+ "3 1 CASH_OUT 181.00 C840083671 181.0 0.00 \n",
+ "4 1 PAYMENT 11668.14 C2048537720 41554.0 29885.86 \n",
+ "\n",
+ " nameDest oldbalanceDest newbalanceDest isFraud isFlaggedFraud \n",
+ "0 M1979787155 0.0 0.0 0 0 \n",
+ "1 M2044282225 0.0 0.0 0 0 \n",
+ "2 C553264065 0.0 0.0 1 0 \n",
+ "3 C38997010 21182.0 0.0 1 0 \n",
+ "4 M1230701703 0.0 0.0 0 0 "
+ ]
+ },
+ "metadata": {
+ "tags": []
+ },
+ "execution_count": 4
+ }
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "tTalrdGXRMMH"
+ },
+ "source": [
+ "# Correcting inconsistency in column name\n",
+ "df = df.rename(columns={'oldbalanceOrg':'oldbalanceOrig'})"
+ ],
+ "execution_count": null,
+ "outputs": []
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "3LzluXf_RMMH"
+ },
+ "source": [
+ "#### Columns are:\n",
+ "* step: A timestamp / date variable that has been made arbitrary for data privacy.\n",
+ "\n",
+ "* type: The type of transaction. type is our only categorical independent variable. Categories include [cash_in, cash_out, debit, payment, transfer].\n",
+ "\n",
+ "* amount: Size of transaction.\n",
+ "\n",
+ "\n",
+ "Next 6 columns are account level info:\n",
+ "\n",
+ "* Suffix *Orig refers to the originating account of the transaction.\n",
+ "\n",
+ "* Suffix *Dest refers to destination account of the transaction.\n",
+ "\n",
+ "* Prefix name* refers to Account ID number. An 'M' prefix, e.g. 'M1979...' denotes a merchant account. The name will not be directly used as a model feature, although we can extract the merchant prefix to create a boolean indicator variable. One thing to note is that balance data is not available for merchants. For merchants, the placeholder value for balances is zero.\n",
+ "\n",
+ "* Prefix oldbalance* refers to account balance before transaction.\n",
+ "\n",
+ "* Prefix newbalance* refers to account balance after transaction (and also great pair of shoes!)\n",
+ "\n",
+ "* isFraud: Our label / dependent variable—whether the transaction was made by a fraudulent agent. 1 for fraudulent, 0 for not fraudulent.\n",
+ "\n",
+ "* isFlaggedFraud: Whether the transaction was flagged as fraud by the \"business model\". Since we are doing our own prediction here, we will ignore this variable\n",
+ "\n",
+ "'NOTE:' - Here Fraudulent means it is a case of Money Laundering or not."
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "fdnSY2waRMMI"
+ },
+ "source": [
+ "### Nulls\n",
+ "\n",
+ "We have no nulls in the table. This will save us the headache of having to fill in or impute missing values, or otherwise exclude incomplete data rows, if required by the model."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "f8Gvuk_ORMMI",
+ "outputId": "3c49e21e-555a-4c12-ccd1-1f127c508db1"
+ },
+ "source": [
+ "df.isnull().values.any()"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "execute_result",
+ "data": {
+ "text/plain": [
+ "False"
+ ]
+ },
+ "metadata": {
+ "tags": []
+ },
+ "execution_count": 6
+ }
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "ZITGINuZRMMK"
+ },
+ "source": [
+ "### Data Imbalance\n",
+ "\n",
+ "As is often the case with fraud data like this, the dataset is highly imbalanced. Over 99.8% of the transaction records are non-fraudulent. Because only a tiny fraction of the dataset represents fraud, fraudulent transactions are likely to be under-represented in our model. If unaddressed, data imbalance can cause issues, such as misleading accuracy metrics in the model. The model may attempt to label every transaction as non-fraudulent, and since any randomly given transaction is highly likely to be non-fraudulent, a high accuracy will be reported, despite incorrectly classifying all fraudulent transactions as non-fraudulent. We should take care to compare model performance metrics considering fraudulent and non-fraudulent data independently."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "r3K1ln6dRMML",
+ "outputId": "22477e3c-9c80-4038-8bcc-0b0569612287"
+ },
+ "source": [
+ "print('The proportion of total transactions labeled as fraudulent is',df['isFraud'].value_counts()[1]/df['isFraud'].size)"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "stream",
+ "text": [
+ "The proportion of total transactions labeled as fraudulent is 0.001290820448180152\n"
+ ],
+ "name": "stdout"
+ }
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "9PXjRLQCRMML",
+ "outputId": "a4975e7c-5561-44dd-a604-91e2050e633e"
+ },
+ "source": [
+ "df['isFraud'].value_counts().plot(kind='bar')"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "execute_result",
+ "data": {
+ "text/plain": [
+ ""
+ ]
+ },
+ "metadata": {
+ "tags": []
+ },
+ "execution_count": 8
+ }
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "akNcS9kaRMMM",
+ "outputId": "0612fccc-8c4c-4368-b2fd-267413b3ad06"
+ },
+ "source": [
+ "df['isFraud'].value_counts()"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "execute_result",
+ "data": {
+ "text/plain": [
+ "0 6354407\n",
+ "1 8213\n",
+ "Name: isFraud, dtype: int64"
+ ]
+ },
+ "metadata": {
+ "tags": []
+ },
+ "execution_count": 9
+ }
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "Z72-qisuRMMM"
+ },
+ "source": [
+ "### Features\n",
+ "\n",
+ "Let's look at frequency plots for the feature variables.\n",
+ "\n",
+ "\n",
+ "#### (i). step\n",
+ "\n",
+ "step is our time variable. Most transactions occur in the first half of our time sample range."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "EIsu0j58RMMM",
+ "outputId": "72900f60-3679-4cbc-ccc0-108d8a2d301a"
+ },
+ "source": [
+ "df['step'].hist()"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "execute_result",
+ "data": {
+ "text/plain": [
+ ""
+ ]
+ },
+ "metadata": {
+ "tags": []
+ },
+ "execution_count": 10
+ },
+ {
+ "output_type": "display_data",
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "
"
+ ]
+ },
+ "metadata": {
+ "tags": [],
+ "needs_background": "light"
+ }
+ }
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "cWlNHR2ZRMMV"
+ },
+ "source": [
+ "The difference is less stark with balancediffDest, and the difference values also tend to be more positive, indicating that the transactions tend to move funds into the destination account."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "lWnKxTtCRMMV",
+ "outputId": "d8859c7a-f2d1-4953-a407-1e4608f29b06"
+ },
+ "source": [
+ "sns.boxplot(x=\"isFraud\", y=\"balancediffDest\", data=df, showfliers=False)"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "execute_result",
+ "data": {
+ "text/plain": [
+ ""
+ ]
+ },
+ "metadata": {
+ "tags": []
+ },
+ "execution_count": 36
+ },
+ {
+ "output_type": "display_data",
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "
"
+ ]
+ },
+ "metadata": {
+ "tags": [],
+ "needs_background": "light"
+ }
+ }
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "A_jid71lRMMV"
+ },
+ "source": [
+ "A commonality in these last two plots was that the balance difference values for isFraud==1 had more variance and spread. This may be because our sample size for fraudulent data is much smaller. By nature, smaller samples are likely to exhibit higher variability."
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "tFHYG-VURMMV"
+ },
+ "source": [
+ "### Preprocessing\n",
+ "\n",
+ "We will now prepare the data for model input. Let's subset the dataframe to only select the variables we will use as model features, as well as the label for classification."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "-BE_TTaaRMMW"
+ },
+ "source": [
+ "features = ['step',\n",
+ " 'type',\n",
+ " 'amount',\n",
+ " 'oldbalanceOrig',\n",
+ " 'newbalanceOrig',\n",
+ " 'oldbalanceDest',\n",
+ " 'newbalanceDest',\n",
+ " 'balancediffOrig',\n",
+ " 'balancediffDest',\n",
+ " 'merchant']\n",
+ "\n",
+ "label = ['isFraud']"
+ ],
+ "execution_count": null,
+ "outputs": []
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "D4_tXkxZRMMW"
+ },
+ "source": [
+ "X = df[features]\n",
+ "y = df[label]"
+ ],
+ "execution_count": null,
+ "outputs": []
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "sCW70FF-RMMW",
+ "outputId": "23e0cebd-f19f-4683-aaf9-7a9760d74f5e"
+ },
+ "source": [
+ "# Before encoding\n",
+ "X.head()"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "execute_result",
+ "data": {
+ "text/html": [
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ "
step
\n",
+ "
type
\n",
+ "
amount
\n",
+ "
oldbalanceOrig
\n",
+ "
newbalanceOrig
\n",
+ "
oldbalanceDest
\n",
+ "
newbalanceDest
\n",
+ "
balancediffOrig
\n",
+ "
balancediffDest
\n",
+ "
merchant
\n",
+ "
\n",
+ " \n",
+ " \n",
+ "
\n",
+ "
0
\n",
+ "
1
\n",
+ "
PAYMENT
\n",
+ "
9839.64
\n",
+ "
170136.0
\n",
+ "
160296.36
\n",
+ "
0.0
\n",
+ "
0.0
\n",
+ "
-9839.64
\n",
+ "
0.0
\n",
+ "
True
\n",
+ "
\n",
+ "
\n",
+ "
1
\n",
+ "
1
\n",
+ "
PAYMENT
\n",
+ "
1864.28
\n",
+ "
21249.0
\n",
+ "
19384.72
\n",
+ "
0.0
\n",
+ "
0.0
\n",
+ "
-1864.28
\n",
+ "
0.0
\n",
+ "
True
\n",
+ "
\n",
+ "
\n",
+ "
2
\n",
+ "
1
\n",
+ "
TRANSFER
\n",
+ "
181.00
\n",
+ "
181.0
\n",
+ "
0.00
\n",
+ "
0.0
\n",
+ "
0.0
\n",
+ "
-181.00
\n",
+ "
0.0
\n",
+ "
False
\n",
+ "
\n",
+ "
\n",
+ "
3
\n",
+ "
1
\n",
+ "
CASH_OUT
\n",
+ "
181.00
\n",
+ "
181.0
\n",
+ "
0.00
\n",
+ "
21182.0
\n",
+ "
0.0
\n",
+ "
-181.00
\n",
+ "
-21182.0
\n",
+ "
False
\n",
+ "
\n",
+ "
\n",
+ "
4
\n",
+ "
1
\n",
+ "
PAYMENT
\n",
+ "
11668.14
\n",
+ "
41554.0
\n",
+ "
29885.86
\n",
+ "
0.0
\n",
+ "
0.0
\n",
+ "
-11668.14
\n",
+ "
0.0
\n",
+ "
True
\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
"
+ ],
+ "text/plain": [
+ " step type amount oldbalanceOrig newbalanceOrig oldbalanceDest \\\n",
+ "0 1 PAYMENT 9839.64 170136.0 160296.36 0.0 \n",
+ "1 1 PAYMENT 1864.28 21249.0 19384.72 0.0 \n",
+ "2 1 TRANSFER 181.00 181.0 0.00 0.0 \n",
+ "3 1 CASH_OUT 181.00 181.0 0.00 21182.0 \n",
+ "4 1 PAYMENT 11668.14 41554.0 29885.86 0.0 \n",
+ "\n",
+ " newbalanceDest balancediffOrig balancediffDest merchant \n",
+ "0 0.0 -9839.64 0.0 True \n",
+ "1 0.0 -1864.28 0.0 True \n",
+ "2 0.0 -181.00 0.0 False \n",
+ "3 0.0 -181.00 -21182.0 False \n",
+ "4 0.0 -11668.14 0.0 True "
+ ]
+ },
+ "metadata": {
+ "tags": []
+ },
+ "execution_count": 39
+ }
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "4ZC6-3pJRMMW"
+ },
+ "source": [
+ "### One-Hot Encoding\n",
+ "\n",
+ "type is a categorical variable, which will need to be encoded as separate boolean dummy variables for each unique value of account type to be used by the model."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "eVefXfQbRMMW",
+ "outputId": "cb50bf80-cb3f-46ca-ef37-415c4a49f4b3"
+ },
+ "source": [
+ "# After encoding (scroll right to see new columns)\n",
+ "X = X.join(pd.get_dummies(X[['type']], prefix='type')).drop(['type'], axis=1)\n",
+ "X.head()"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "execute_result",
+ "data": {
+ "text/html": [
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ "
step
\n",
+ "
amount
\n",
+ "
oldbalanceOrig
\n",
+ "
newbalanceOrig
\n",
+ "
oldbalanceDest
\n",
+ "
newbalanceDest
\n",
+ "
balancediffOrig
\n",
+ "
balancediffDest
\n",
+ "
merchant
\n",
+ "
type_CASH_IN
\n",
+ "
type_CASH_OUT
\n",
+ "
type_DEBIT
\n",
+ "
type_PAYMENT
\n",
+ "
type_TRANSFER
\n",
+ "
\n",
+ " \n",
+ " \n",
+ "
\n",
+ "
0
\n",
+ "
1
\n",
+ "
9839.64
\n",
+ "
170136.0
\n",
+ "
160296.36
\n",
+ "
0.0
\n",
+ "
0.0
\n",
+ "
-9839.64
\n",
+ "
0.0
\n",
+ "
True
\n",
+ "
0
\n",
+ "
0
\n",
+ "
0
\n",
+ "
1
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
1
\n",
+ "
1
\n",
+ "
1864.28
\n",
+ "
21249.0
\n",
+ "
19384.72
\n",
+ "
0.0
\n",
+ "
0.0
\n",
+ "
-1864.28
\n",
+ "
0.0
\n",
+ "
True
\n",
+ "
0
\n",
+ "
0
\n",
+ "
0
\n",
+ "
1
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
2
\n",
+ "
1
\n",
+ "
181.00
\n",
+ "
181.0
\n",
+ "
0.00
\n",
+ "
0.0
\n",
+ "
0.0
\n",
+ "
-181.00
\n",
+ "
0.0
\n",
+ "
False
\n",
+ "
0
\n",
+ "
0
\n",
+ "
0
\n",
+ "
0
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
3
\n",
+ "
1
\n",
+ "
181.00
\n",
+ "
181.0
\n",
+ "
0.00
\n",
+ "
21182.0
\n",
+ "
0.0
\n",
+ "
-181.00
\n",
+ "
-21182.0
\n",
+ "
False
\n",
+ "
0
\n",
+ "
1
\n",
+ "
0
\n",
+ "
0
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
4
\n",
+ "
1
\n",
+ "
11668.14
\n",
+ "
41554.0
\n",
+ "
29885.86
\n",
+ "
0.0
\n",
+ "
0.0
\n",
+ "
-11668.14
\n",
+ "
0.0
\n",
+ "
True
\n",
+ "
0
\n",
+ "
0
\n",
+ "
0
\n",
+ "
1
\n",
+ "
0
\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
"
+ ],
+ "text/plain": [
+ " step amount oldbalanceOrig newbalanceOrig oldbalanceDest \\\n",
+ "0 1 9839.64 170136.0 160296.36 0.0 \n",
+ "1 1 1864.28 21249.0 19384.72 0.0 \n",
+ "2 1 181.00 181.0 0.00 0.0 \n",
+ "3 1 181.00 181.0 0.00 21182.0 \n",
+ "4 1 11668.14 41554.0 29885.86 0.0 \n",
+ "\n",
+ " newbalanceDest balancediffOrig balancediffDest merchant type_CASH_IN \\\n",
+ "0 0.0 -9839.64 0.0 True 0 \n",
+ "1 0.0 -1864.28 0.0 True 0 \n",
+ "2 0.0 -181.00 0.0 False 0 \n",
+ "3 0.0 -181.00 -21182.0 False 0 \n",
+ "4 0.0 -11668.14 0.0 True 0 \n",
+ "\n",
+ " type_CASH_OUT type_DEBIT type_PAYMENT type_TRANSFER \n",
+ "0 0 0 1 0 \n",
+ "1 0 0 1 0 \n",
+ "2 0 0 0 1 \n",
+ "3 1 0 0 0 \n",
+ "4 0 0 1 0 "
+ ]
+ },
+ "metadata": {
+ "tags": []
+ },
+ "execution_count": 40
+ }
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "1fUyKwo7RMMX"
+ },
+ "source": [
+ "### Modeling\n",
+ "\n",
+ "We will first split the data into training and test sets, with a 70/30 split. This is done to measure and avoid possible overfitting, an issue in which the model is too specific to the dataset it trains on, and is not as generalizable to out-of-sample data. By measuring the model's performance on separate data, we have a more fair assessment of the model's ability to extrapolate its predictions to new or unseen information."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "0z_wDtbRRMMX"
+ },
+ "source": [
+ "from sklearn.model_selection import train_test_split "
+ ],
+ "execution_count": null,
+ "outputs": []
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "U0-tJdkHRMMX",
+ "outputId": "2d8099c8-15c6-452b-db06-6791fdd085d8"
+ },
+ "source": [
+ "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=1)\n",
+ "print(X_train.shape)\n",
+ "print(y_train.shape)\n",
+ "print(X_test.shape)\n",
+ "print(y_test.shape)"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "stream",
+ "text": [
+ "(4453834, 14)\n",
+ "(4453834, 1)\n",
+ "(1908786, 14)\n",
+ "(1908786, 1)\n"
+ ],
+ "name": "stdout"
+ }
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "ijbCSU3wRMMX"
+ },
+ "source": [
+ "#### Decision Tree\n",
+ "\n",
+ "At a high level, a decision tree model works by sequentially splitting the dataset along value thresholds of the predictor variables. The splits are determined by a criteria that quantifies how different two sets of data are (by default, Gini impurity). For example, if fraudulent transactions most frequently occur with amounts over 1,000,000, while non-fraudulent transactions are more frequently of lower amounts, our decision tree is likely to use amount as a node that splits our dataset into transactions of those over 1,000,000 and those that are less. The decision tree then considers each of those two groups and attempts to perform another split on one of the other features, and so on, until the split groups have zero impurity (all the members of each group have the same classification). While in actuality the data does not contain such explicit separations, this is the general logic used by the algorithm. In a basic sense, it can be thought of as categorizing the data based on its features. It repeats the process iteratively on each branch until an end (leaf) is reached, hence the \"tree\".\n",
+ "\n",
+ "An advantage of using decision trees is that the results are highly interpretable and easy to visualize. This means we will be able to see exactly how the decision tree constructs itself, and what features it prioritizes in its construction. It will show us how the data is being split and by what values for each feature will lead to a certain classification. The same cannot be said for certain \"black box\" methods of modeling, such as neural networks—which despite their potential for high performance, are often too complex to be intuitively understood and explained, even by the model developer.\n",
+ "\n",
+ "Let's implement the decision tree and measure its performance."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "xUmvpRnHRMMX"
+ },
+ "source": [
+ "from sklearn.tree import DecisionTreeClassifier"
+ ],
+ "execution_count": null,
+ "outputs": []
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "tfk0tF-hRMMY"
+ },
+ "source": [
+ "dt_clf = DecisionTreeClassifier(random_state=1)\n",
+ "\n",
+ "dt_clf = dt_clf.fit(X_train,y_train)\n",
+ "\n",
+ "y_pred = dt_clf.predict(X_test)"
+ ],
+ "execution_count": null,
+ "outputs": []
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "wNpGhHfQRMMY"
+ },
+ "source": [
+ "Now that the model has been fit, we can create predictions for the test set and see how they compare to the actual data. A sample of the predictions can be seen below."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "PVDaQbNlRMMY",
+ "outputId": "2b07b3a3-6fc8-48da-a171-59c7fe9eac4a"
+ },
+ "source": [
+ "result = pd.DataFrame({'actual':y_test['isFraud'], 'predicted':y_pred})\n",
+ "result[result['actual']==1]"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "execute_result",
+ "data": {
+ "text/html": [
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ "
actual
\n",
+ "
predicted
\n",
+ "
\n",
+ " \n",
+ " \n",
+ "
\n",
+ "
637351
\n",
+ "
1
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
3434399
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
3137121
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
1069777
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6259961
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6166140
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6065199
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
2571515
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6261323
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6027400
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
1325512
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
1978760
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
1059545
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
5563207
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
4388723
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
1237397
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
1832574
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6281815
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6010861
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6118824
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6259921
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
5987920
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
3362343
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
4945978
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
1030682
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
4892274
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
5762775
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
5994349
\n",
+ "
1
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
5987525
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
5990452
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
...
\n",
+ "
...
\n",
+ "
...
\n",
+ "
\n",
+ "
\n",
+ "
6118269
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6187204
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
782395
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
3913781
\n",
+ "
1
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
5956511
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
651948
\n",
+ "
1
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
1059497
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
3182792
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
3610372
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
1030639
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
5188014
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
1709833
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
4263779
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6060437
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
2568139
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6081340
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6051520
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6006723
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6040734
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
6168735
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
1059706
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
562242
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
2796961
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
3610999
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
5823885
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
2214171
\n",
+ "
1
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
1502704
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
1030672
\n",
+ "
1
\n",
+ "
0
\n",
+ "
\n",
+ "
\n",
+ "
5407591
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ "
\n",
+ "
3960261
\n",
+ "
1
\n",
+ "
1
\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
2468 rows × 2 columns
\n",
+ "
"
+ ],
+ "text/plain": [
+ " actual predicted\n",
+ "637351 1 0\n",
+ "3434399 1 1\n",
+ "3137121 1 1\n",
+ "1069777 1 1\n",
+ "6259961 1 1\n",
+ "6166140 1 1\n",
+ "6065199 1 1\n",
+ "2571515 1 1\n",
+ "6261323 1 1\n",
+ "6027400 1 1\n",
+ "1325512 1 1\n",
+ "1978760 1 1\n",
+ "1059545 1 1\n",
+ "5563207 1 1\n",
+ "4388723 1 1\n",
+ "1237397 1 1\n",
+ "1832574 1 1\n",
+ "6281815 1 1\n",
+ "6010861 1 1\n",
+ "6118824 1 1\n",
+ "6259921 1 1\n",
+ "5987920 1 1\n",
+ "3362343 1 1\n",
+ "4945978 1 1\n",
+ "1030682 1 1\n",
+ "4892274 1 1\n",
+ "5762775 1 1\n",
+ "5994349 1 0\n",
+ "5987525 1 1\n",
+ "5990452 1 1\n",
+ "... ... ...\n",
+ "6118269 1 1\n",
+ "6187204 1 1\n",
+ "782395 1 1\n",
+ "3913781 1 0\n",
+ "5956511 1 1\n",
+ "651948 1 0\n",
+ "1059497 1 1\n",
+ "3182792 1 1\n",
+ "3610372 1 1\n",
+ "1030639 1 1\n",
+ "5188014 1 1\n",
+ "1709833 1 1\n",
+ "4263779 1 1\n",
+ "6060437 1 1\n",
+ "2568139 1 1\n",
+ "6081340 1 1\n",
+ "6051520 1 1\n",
+ "6006723 1 1\n",
+ "6040734 1 1\n",
+ "6168735 1 1\n",
+ "1059706 1 1\n",
+ "562242 1 1\n",
+ "2796961 1 1\n",
+ "3610999 1 1\n",
+ "5823885 1 1\n",
+ "2214171 1 0\n",
+ "1502704 1 1\n",
+ "1030672 1 0\n",
+ "5407591 1 1\n",
+ "3960261 1 1\n",
+ "\n",
+ "[2468 rows x 2 columns]"
+ ]
+ },
+ "metadata": {
+ "tags": []
+ },
+ "execution_count": 45
+ }
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "evJ2gfc0RMMY"
+ },
+ "source": [
+ "Let's look at some metrics commonly used for measuring model performance. The sklearn.metrics.classification_report function gives us a nice printout for assessing performance, including precision, recall, and f1-score for each of the classes independently. The separation for the classes is important especially when dealing with imbalanced data, as we are now. All of these metrics range from 0 to 1, higher numbers representing better accuracy.\n",
+ "\n",
+ "* Precision: The proportion of values selected by the model that should be selected. Precision is penalized by having more false positives.\n",
+ "* Recall: The proportion of values that should be selected, that are actually selected by the model. Recall is penalized by having more false negatives.\n",
+ "* F1 score: The harmonic mean of precision and recall.\n",
+ "\n",
+ "Another measure we can use is Area Under Curve (AUC), which refers to the area under the ROC curve, which is a plot of the true positive rate against the false positive rate. AUC also ranges from 0 to 1 and is frequently used as a quick way to compare model performance."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "1xYBoqQGRMMY"
+ },
+ "source": [
+ "from sklearn import metrics "
+ ],
+ "execution_count": null,
+ "outputs": []
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "hkTGOyBoRMMZ",
+ "outputId": "9d889027-41ae-48f0-ad81-04c3bec94dde"
+ },
+ "source": [
+ "print(metrics.classification_report(y_test, y_pred))\n",
+ "\n",
+ "fpr, tpr, thresholds = metrics.roc_curve(y_test, y_pred)\n",
+ "print('AUC:', metrics.auc(fpr, tpr))"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "stream",
+ "text": [
+ " precision recall f1-score support\n",
+ "\n",
+ " 0 1.00 1.00 1.00 1906318\n",
+ " 1 0.88 0.87 0.88 2468\n",
+ "\n",
+ " accuracy 1.00 1908786\n",
+ " macro avg 0.94 0.94 0.94 1908786\n",
+ "weighted avg 1.00 1.00 1.00 1908786\n",
+ "\n",
+ "AUC: 0.9371168999215428\n"
+ ],
+ "name": "stdout"
+ }
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "pXA4MBo7RMMZ",
+ "outputId": "ec5459a4-59c2-49a1-f0d0-9cddfc4ee529"
+ },
+ "source": [
+ "from sklearn.tree import plot_tree\n",
+ "fig = plt.figure(figsize=(17,10))\n",
+ "_ = plot_tree(dt_clf, max_depth=2, feature_names=list(X_train.columns), filled=True, fontsize=12)"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "display_data",
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ "
"
+ ]
+ },
+ "metadata": {
+ "tags": [],
+ "needs_background": "light"
+ }
+ }
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "RdRy8FIqRMMZ"
+ },
+ "source": [
+ "In order to objectively determine each features' influence on the predictions, we can calculate, summed up for each feature, the amount by which it reduces impurity in each node it represents. Sklearn provides these values in the feature_importances_ attribute of DecisionTreeClassifier."
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "wjUiBsQeRMMZ",
+ "outputId": "7a028f0d-b31b-4d9c-8960-c8db0936cc92"
+ },
+ "source": [
+ "fi = pd.DataFrame({'features':X_train.columns,'importance':dt_clf.feature_importances_}).sort_values(by=['importance'], ascending=False)\n",
+ "fi"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "execute_result",
+ "data": {
+ "text/html": [
+ "
\n",
+ "\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
\n",
+ "
features
\n",
+ "
importance
\n",
+ "
\n",
+ " \n",
+ " \n",
+ "
\n",
+ "
6
\n",
+ "
balancediffOrig
\n",
+ "
0.405893
\n",
+ "
\n",
+ "
\n",
+ "
5
\n",
+ "
newbalanceDest
\n",
+ "
0.152945
\n",
+ "
\n",
+ "
\n",
+ "
7
\n",
+ "
balancediffDest
\n",
+ "
0.091866
\n",
+ "
\n",
+ "
\n",
+ "
13
\n",
+ "
type_TRANSFER
\n",
+ "
0.086673
\n",
+ "
\n",
+ "
\n",
+ "
1
\n",
+ "
amount
\n",
+ "
0.081542
\n",
+ "
\n",
+ "
\n",
+ "
0
\n",
+ "
step
\n",
+ "
0.063562
\n",
+ "
\n",
+ "
\n",
+ "
2
\n",
+ "
oldbalanceOrig
\n",
+ "
0.048047
\n",
+ "
\n",
+ "
\n",
+ "
3
\n",
+ "
newbalanceOrig
\n",
+ "
0.029842
\n",
+ "
\n",
+ "
\n",
+ "
10
\n",
+ "
type_CASH_OUT
\n",
+ "
0.020618
\n",
+ "
\n",
+ "
\n",
+ "
4
\n",
+ "
oldbalanceDest
\n",
+ "
0.019012
\n",
+ "
\n",
+ "
\n",
+ "
8
\n",
+ "
merchant
\n",
+ "
0.000000
\n",
+ "
\n",
+ "
\n",
+ "
9
\n",
+ "
type_CASH_IN
\n",
+ "
0.000000
\n",
+ "
\n",
+ "
\n",
+ "
11
\n",
+ "
type_DEBIT
\n",
+ "
0.000000
\n",
+ "
\n",
+ "
\n",
+ "
12
\n",
+ "
type_PAYMENT
\n",
+ "
0.000000
\n",
+ "
\n",
+ " \n",
+ "
\n",
+ "
"
+ ],
+ "text/plain": [
+ " features importance\n",
+ "6 balancediffOrig 0.405893\n",
+ "5 newbalanceDest 0.152945\n",
+ "7 balancediffDest 0.091866\n",
+ "13 type_TRANSFER 0.086673\n",
+ "1 amount 0.081542\n",
+ "0 step 0.063562\n",
+ "2 oldbalanceOrig 0.048047\n",
+ "3 newbalanceOrig 0.029842\n",
+ "10 type_CASH_OUT 0.020618\n",
+ "4 oldbalanceDest 0.019012\n",
+ "8 merchant 0.000000\n",
+ "9 type_CASH_IN 0.000000\n",
+ "11 type_DEBIT 0.000000\n",
+ "12 type_PAYMENT 0.000000"
+ ]
+ },
+ "metadata": {
+ "tags": []
+ },
+ "execution_count": 50
+ }
+ ]
+ },
+ {
+ "cell_type": "code",
+ "metadata": {
+ "id": "Ait2n1MjRMMZ",
+ "outputId": "a6e59f6c-c9ad-46c9-d7ab-4445774ca9da"
+ },
+ "source": [
+ "sns.barplot(x=\"importance\", y=\"features\", data=fi)"
+ ],
+ "execution_count": null,
+ "outputs": [
+ {
+ "output_type": "execute_result",
+ "data": {
+ "text/plain": [
+ ""
+ ]
+ },
+ "metadata": {
+ "tags": []
+ },
+ "execution_count": 51
+ },
+ {
+ "output_type": "display_data",
+ "data": {
+ "image/png": "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\n",
+ "text/plain": [
+ ""
+ ]
+ },
+ "metadata": {
+ "tags": [],
+ "needs_background": "light"
+ }
+ }
+ ]
+ },
+ {
+ "cell_type": "markdown",
+ "metadata": {
+ "id": "2gx8XH-BRMMa"
+ },
+ "source": [
+ "### Potential Improvements\n",
+ "We have established a basic process for exploratory data analysis and modeling that has resulted in an acceptable predictor for our transaction data. While I will end our work here, there are other modeling techniques that we could try, such as SVM and Logistic Regression. There are also additional techniques to increase model performance, as well as better ways to measure performance, such as hyper-parameter tuning and cross-fold validation. It would be prudent to use these methods if we were tasked with informing business decisions that could affect large amounts of revenue or costs"
+ ]
+ }
+ ]
+}
\ No newline at end of file
![\"Open](\"https://colab.research.google.com/assets/colab-badge.svg\"){kind=icon}