Portfolio Part $3-$ Analysis of Mobile Price Data $(2024$ S$1)$

In this Portfolio task, you will work on a new dataset named 'Mobile Price Data', it contains numerous details about mobile phone hardware, specifications, and prices. Your main task is to train classification models to predict mobile phone prices $(\text{'}$price range' in the dataset$)$and evaluate the strengths and weaknesses of these models.

Here's the explanation of each column:

Column Meaning

battery power Total energy a battery can store in one time measured in mAh

blue Has bluetooth or not

clock speed speed at which microprocessor executes instructions

dual sim Has dual sim support or not

fc Front Camera mega pixels

four g Has $4$G or not

int memory Internal Memory in Gigabytes

m dep Mobile Depth in cm

mobile wt Weight of mobile phone

n cores Number of cores of processor

pc Primary Camera mega pixels

px height Pixel Resolution Height

px width Pixel Resolution Width

ram Random Access Memory in Mega Bytes

sc h Screen Height of mobile in cm

sc w Screen Width of mobile in cm

talk time longest time that a single battery charge will last when you are

three g Has $3$G or not

touch screen Has touch screen or not

wifi Has wifi or not

price range This is the target variable with value of $0($low cost$),1($medium cost$),2($high cost$)$ and $3($very high cost$)$

Blue, dual sim, four g$,$ three g$,$ touch screen, and wifi are all binary attributes, $0$ for not and $1$ for yes.

Your high level goal in this notebook is to build and evaluate predictive models for 'price range' from other available features. More specifically, you need to complete the following major steps:

Explore the data and clean the data if necessary. For example, remove abnormal instanaces and replace missing values.

Study the correlation between 'price range' with other features. And select the variables that you think are helpful for predicting the price range. We do not limit the number of variables.

Split the dataset $($Trainging set : Test set $=8$ : $2)$

Train a logistic regression model to predict 'price range' based on the selected features $($from the second step$).$ Calculate the accuracy of your model. $($You are required to report the accuracy from both training set and test set.$)$ Explain your model and evaluate its performance $($Is the model performing well? If yes, what factors might be contributing to the good performance of your model? If not, how can improvements be made?$).$

Train a KNN model to predict 'price range' based on the selected features $($you can use the features selected from the second step and set K with an ad$-$hoc manner in this step. Calculate the accuracy of your model. $($You are required to report the accuracy from both training set and test set.$)$

Tune the hyper$-$parameter K in KNN $($Hints: GridsearchCV$),$ visualize the results, and explain how K influences the prediction performance.