Q$4.$ Support Vector Machines $($SVM$)$ is a supervised learning algorithm, which can be applicable to both classification and regression. The data set provided contains normal and fraudulent transactions in excel file $$Week$6-$Fraud$\_$data$$ in sheets $$FraudTrain$$ and $$FraudTest$.$ Using Support Machine Model $($SVM$),$ predict whether the transactions are Normal or Fraudulent based on the features of the transactions in the dataset.

Use is$\_$fraud as target variable and features as independent variables. Convert all textual categorical variables to numeric, and clean data if necessary. For SVM model, add Vaimal Machine Learning Add$-$in$.$

$($Note: Install Vaimal Add$-$in attached with the assignment using installation instructions below and detailed instruction in the Manual.$)$

Please follow below process for model development:

$1.$ Import or load Data: Place data in an Excel worksheet.

$2.$ Perform Data Preprocessing: Deal with missing data, data normalization, and encoding categorical inputs.

$($Vaimal has several utilities for preprocessing data such as Data Manager.$)$

$3.$ Select a Model: To use and design it$.$ Select which model to use and the design parameters.

$4.$ Train the Model: Using training data with known outputs, train the model.

$5.$ Test the Model Using different data than the training data, test the model$$s ability to predict versus known outputs.

$6.$ Prediction: Use the model to make predictions of data with unknown output.

Input Variables:

trans$\_$date$\_$trans$\_$time, cc$\_$num, first, last, merchant, category, amt, gender, street, city, state, zip, lat, long, city$\_$pop, job, unix$\_$time, merch$\_$lat, and merch$\_$long. Please convert categorical variables to Numbers such as Gender, and category etc. You can drop unwanted columns and use below columns as INPUT columns:

Output Variables:

Please use column $$is$\_$fraud$$ as a target variable.

Data Flag for Training or Testing:

The column $$Data$\_$FLAG$$ differentiates the data as Training and Testing. Please supply input to the model as per inputs.

a$.$ Provide descriptive measures of column $$amt$$: count, Min, Max, Mean, and Std$.$ deviation. Write the count of fraudulent and normal transactions.

b$.$ Show graph of Normalized frequency for columns category and Is$\_$fraud to show frequency for normal and fraudulent transactions.

c$.$ Perform Error analysis using confusion matrix, which is created with four categories $-$ true positives $($TP$),$ false positives $($FP$),$ true negatives $($TN$),$ and false negatives $($FN$).$

Precision $($P$)=$ TP$/($TP $+$ FP$)$

Recall $($R$)=$ TP$/($TP $+$ FN$)$

Accuracy $=($TP $+$ TN$)/($TP $+$ FP $+$ TN $+$ FN$)$

F$1$ score $=(2*$ P $*$ R$\backslash /($P $+$ R$)$

d$.$ Based on the features below, predict the status of the transaction if it is Fraud:

trans$\_$date$\_$trans$\_$time: $12/2/202022$:$27$

cc$\_$num: $3\mathrm{.}588$E$+15$

merchant: fraud$\_$Torphy$-$Goyette

category: shopping$\_$pos

amt: $1318\mathrm{.}89$

first: Jason, last: Johnson, gender: M

street: $5942$ Thomas Park, city: Craig

state: AK$,$ zip: $99921$

lat: $55\mathrm{.}4732,$ long: $-133\mathrm{.}1171,$ city$\_$pop: $1920$

job: Commissioning editor, dob: $6/17/1997$

trans$\_$num: $2682$f$81$f$3$f$9$e$070$b$7$abc$721$ca$4$bd$5862$

unix$\_$time: $1386023256$

merch$\_$lat: $54\mathrm{.}801713,$ merch$\_$long: $-133\mathrm{.}669108$

is$\_$fraud: $1$