README.md

OS_CA-3-MultiThread-price-classifier

An OS course project to predict price of phones by train data and weight Matrix

this project has two implementation:

serial

parallel

How to run ?

For serial part you in serial directory just run make command to make project by g++. then run executable file in this form :

‫‪./PhonePricePrediction.out‬‬‬ <datasets‬‬ path>

like this :

‫‪./PhonePricePrediction.out‬‬ ‫‪datasets/

And for parallel part you first copy breaker.sh scripts to your dataset directory near train.csv and weight.csv files and run it in this form:

‫‪./breaker.sh‬‬‬ <Number of split files>

that this number must be equal to thread number that defined in main.cpp Default in this project like this :

./breaker.sh 6

Then in parallel directory just run make command to make project by g++.

‫‪./PhonePricePrediction.out‬‬‬ <datasets‬‬ path>

like this :

‫‪./PhonePricePrediction.out‬‬ ‫‪datasets/

note : you can change num of thread by edit NUMBER_OF_THREADS define to your custom value in main.cpp file

desire output format :

‫‪Accuracy:‬‬ ‫‪<accuracy_percentage>%‬‬

like this :

‫‪Accuracy:‬‬ ‫‪93.05%‬‬

dataset format :

in dataset directory you must have two files: train.csv file weights.csv file that train had entries of features and weights file is weight vector for each price range class [csv files must have header.]

Created by Amin Fazeli on 16 Dec 2020.

OS Course - Fall 1399 - University of Tehran