MACHINE LEARNING LECTURE PROJECT

a) Find a dataset from Internet with at least 10.000 data in it.

Kaggle https://www.kaggle.com/datasets

UCI Machine Learning Repository https://archive.ics.uci.edu/ml/datasets.php

https://www.v7labs.com/blog/best-free-datasets-for-machine-learning

https://imerit.net/blog/the-60-best-free-datasets-for-machine-learning-all-pbm/

CMU

b) Show related information about the dataset. (How many records does it have? What are the features? Types of the features?.... etc.)

Google Dataset Search:

CMU Libraries: Discover high-quality datasets thanks to the collection of Huajin Wang, at

Dataset should contain at least 15 features in it. DATASET NAME :

DATASET WEB LINK :

DATASET INFO

How many records does it contains

How many features does it have

How many different classes exist in the dataset?

What are number of examples for each classes?

How many NULL values exist? (depending on the features distinctly) Which features are not numeric?

c) Use Label Encoding for at least one of the features (Explain your reason why do make this operation?)

*****

*****

*****

d) Use One Hot encoding for at least one of the features (Explain your reason why do make this operation?)

*****

*****

*****

e) Analyze the Missing Values

a. Delete some columns (Explain your reason why do make this operation?)

*****

*****

*****

b. Delete some rows (Explain your reason why do make this operation?)

*****

*****

*****

c. Impute some missing data (Explain your reason why do make this operation?)

*****

*****

*****

f) FindthebestcorrelatedFeaturesintheDataset DISPLAY THE CORRELATION CALCULATION *****

*****

*****

g) Execute a Normalization/Scaling in the Dataset

PUT THE SCREENSHOT OF DATA.HEAD BEFORE AND AFTER THE OPERATION

*****

*****

*****

h) Train your new dataset at least 5 different Machine Learning algorithms

THE PREFERRED ALGORITHMS ARE *****

*****

i) Use5-foldapproachtomeasuretheperformanceofthesystem

WITH A RANDOM SELECTION WE REACHED THE FOLLOWING RESULTS *****

WITH 5-FOLD APPROACH I REACHED THE FOLLOWING RESULT

*****

j) Puttheirresultstoatabletomakeacomparison

*****

*****

*****

k) Calculate the training time for all of them

*****

*****

*****

l) Selectthebest10featuresfromthedatabase

SHOW THE LIST OF THE FEATURES *****

*****

*****

m) Write a Conference paper to Show all your reached results.

a) Find a dataset frum Internet with at least 10.000 data in it. - Kagole htarsi//rowoukapale.comidatasets - UCi Machine Leaming Repository hetos/ifarchive, ics.veledu/m /efatescts pho - htepsid/ment,net/bog/the-49-best-frew-datavets-for-mechine-learning-al-gbm/' - Google Datirect Searci: - CMJ Lararies Discover high-quality dotasets thanks to the collection of thajin wang, at: CMJ b) Show rebated informacion about ehe dacaset. (tow many recorts does it nevel what, are the features? Types of the features?a. etc.) [15 Points) - Dataset should contan at least 15 feahures in it. DATAIT wast gATAST WIE BA. DATAYT IAP Hew many rexwes des a contains How many tatares does it has Pew muny aftereve aleves aid in the tataie? What wer sumber al eandess sor each elwsie? What teathen art ad nemere? c) Use Label Encoding for at least one of the features (Explais your reason "why do make this operation?") (10 Points) d) Use One Hot encoding for at least one of the features fExplain your mason "why de make this eperation?"] (10 Pointal) cosec entew t1 Analyze the Missing Values a. Dellete some columns [Explain powr neason "why do maiz this operabion?7) itg Points) b. Delete some rows (Explain your reason "why do make this uperation)? \{10 Poinks) c. Impute some missing data (Explain your reason "ahy do maie thes eperetion?7) (to Peinti) 1) Find the best conriated Festutes in the bataset (10 Points) DISPLAY Tod CoegeLATICN CALCULATION 9]. Exeoube Normalization/scoling in the Doeater iH: DisQatict