Machine Learning Analysis of Wisconsin and Mnist Datasets

Running head: MACHINE LEARNING
Machine Learning
Name of the Student:
Name of the University:
University Roll:

Machine Learning2
Executive Summary
The main objective of this report is to analyse two different datasets with variety of machine learning
algorithms. Here the datasets used are Wisconsin and mnist dataset. Here L1 and L2 regularisation is used to
find the accuracy of the model. Various graphs and tables have been included to process the required result.
Logistic Regression has been performed to measure the accuracy of different models developed. Further
accuracy also depends of hardware specification. The first part concludes the Binary classification of
Wisconsin dataset and then performed Multiclass classification of mnist dataset. There have been different
graph and tables were shown for different dataset to analyse the datasets. At the end we discussed about the
conclusion of the report topic.

Machine Learning3
Table of Content
Introduction.........................................................................................................................................................4
Discussion............................................................................................................................................................5
Wisconsin Dataset...........................................................................................................................................5
Binary Classification...................................................................................................................................5
Mnist Dataset.................................................................................................................................................10
Multiclass Classification...........................................................................................................................10
Over fitting.................................................................................................................................................11
Under fitting...............................................................................................................................................12
Conclusion.........................................................................................................................................................14
References..........................................................................................................................................................15

Machine Learning4
Introduction
Machine learning technically and statistically a subpart of Artificial intelligence. The purpose of
machine learning is to understand the structures of the data and fit that particular data into appropriate models
that can be understood and utilized by normal people (Marsland 2014).
Logistic regression in general a classification algorithm that is used to do observation over discrete
set of classes for a dataset. Some classification problem examples are, is the email is spam or not, fraud cases
in online transactions or not, Tumor Malignant or Benign (Alpaydin 2014). Logistic regression has the
capability to completely transform its output using the logistic sigmoid function which always returns a
probability value. Logistic regression is off two types which will be discussed later on while going deep into
the project (Witten et al. 2016).
Nowadays machine learning is been used everywhere from automation and appliances to giving
intelligent solutions, industries from every sector trying to get benefits from it (Pedregosa et al. 2011). We are
already been using such devices that utilizes this technology (Harrington 2012). An important example is, a
fitness tracker like Mi Band, or some intelligent home appliances like Amazon echo. Though there are huge
numbers are gadgets that are recently runs ML algorithms to get the best out of it. (Snoek et al. 2012).
In short it can be said that the main advantage of ML is to get prediction analysis.
Machine learning plays a crucial role to prediction systems. Let consider an example to compute the
probability of a fault case, thus the model will need to classify the available data in groups so that the model
can work properly.
Image recognition is one of the general purpose of machine Learning is face detection or object detection.
The category is separated for each person in a separate database of several people.
Speech Recognition is used in voice searches and more. It also translates spoken words to text format.
With speech recognition one can control home appliances, gadgets and even make calls through speech.
This can also be helpful in data entry filed.
Machine learning plays a crucial role in Medical diagnose as in this situation algorithms are trained to find
out cancer tissues and also if a person have tumor or not.
Financial and banking sectors also depend on Machine learning algorithms for their benefits like trading,
loan and many others sources.