UltimateAnalyticsBenchmark

To get a score on system performace of your machine in the field of analytics

#Running the benchmark

• Clone or Download this repo
• cd to the cloned or Downloaded(Might need to unzip if downloaded) repo
• issue python run.py from commandline

If running this code on bash on ubuntu on windows, you need to do export KMP_AFFINITY=disabled

#Requirements

• Python 3.5 or above
• statsmodels
• pandas
• numpy-mkl

**Best way to run this is install Anaconda 3 from Continuum Analytics and run it **

Some Scores: ####System 1:

1. OS: Ubuntu
2. Processor: intel i5-5th gen MQ processor
3. Ram:- 8 GB DDR3
4. System Type: Desktop
5. Score : 5.6

####System 2:

1. OS: Windows 10
2. Processor: intel i5-6th gen U processor
3. Ram: 8 GB
4. System Type: Laptop
5. Score: 2.5

####System 3:

1. OS: Ubuntu on Windows 10 (After anniversary update)
2. Processor: intel i5-5th gen U processor
3. Ram: 12 GB
4. System Type: Laptop
5. Score: 2.15

####System 4:

1. OS: Windows 10
2. Processor: intel i5-5th gen U processor
3. Ram: 12 GB
4. System Type: Laptop
5. Score: 2.0

####System 5:

1. OS: Windows 10
2. Processor: intel i3-5th gen U processor
3. Ram: 4 GB
4. System Type: Laptop
5. Score: 1.3

#Introduction

##Methodology It works by calculating various statistical values and models (explained later) and compares with a base and it uses Hyperbolic curve formulae which highly penalizes slow machines and at the same time highly rewards good machines.

##Score Range Theoratically score will range from [0, infi)

##Python version Built using python 3

#Functions used

##Generating list Using numpy to generate a very large list of random floats

The folwing functions are run several times and score is calculated

##Sorting the List Using self implemented quicksort to sort.

##Min Finding the minimum element of the list.

##Max Finding the max element of the lis.

##Mean Finding the mean of the list.

##Standard Deviation Finding the sample standard deviation of the list

##Kurtosis Finding the excess Kurtosis of the list

##Pandas Analysis and Training several models The sample data in the "Data/data.xlsx" file is used as the sample data and various operations are performed on the data and the following models are trained: ###1)Ordinary Least Sqaure ###2)Weighted Least Square ###3)GLSAR (Generalized least square with AR(p) convergence) ###4)Robust Linear Model

Except model creation, all methods are self implemented