Scribd
Upload
100 views

Roadmap To Build A Machine Learning Model

This document outlines the key steps to build a machine learning model: 1) Understand the basics of math, stats, and machine learning concepts. 2) Clean and pre-process raw data to remove errors and structure the data. 3) Identify important features in the data and select or engineer new features. 4) Divide the cleaned data into training, validation, and testing sets. 5) Select an appropriate machine learning algorithm based on the type of problem. 6) Build and evaluate the model using the training and validation sets.

Uploaded by

Enock
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
100 views

Roadmap To Build A Machine Learning Model

This document outlines the key steps to build a machine learning model: 1) Understand the basics of math, stats, and machine learning concepts. 2) Clean and pre-process raw data to remove errors and structure the data. 3) Identify important features in the data and select or engineer new features. 4) Divide the cleaned data into training, validation, and testing sets. 5) Select an appropriate machine learning algorithm based on the type of problem. 6) Build and evaluate the model using the training and validation sets.

Uploaded by

Enock
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

BY AKSHAY TOSHNIWAL

ROADMAP TO
BUILD A
MACHINE
LEARNING
MODEL
A complete guide to build machine learning
system (an area under artificial intelligence)
1 Roadmap to build a machine learning model

UNDERSTAND THE BASICS

There are machine learning libraries available


that will ease your work, but you need to focus
on the basics before you understand them.

Understand the basics of math and stats that


include topics like, linear algebra, calculus,
probability, and statistics.
2 Roadmap to build a machine learning model

DATA PRE-PROCESSING

Real-time/real-world data is dirty & noisy and it


contains a lot of outliers (unexpected data
records) that are not relevant for the ML
modelling.

Cleaning this is important to avoid any


unexpected results and also to structure the
data.
3 Roadmap to build a machine learning model

IDENTIFYING FEATURE

Any given dataset has multiple features and


using these features the machine learning
model is designed.

Two ways to do this, selecting the features


(feature selection) or creating new features
(feature engineering). Relevant feature
identification can make the model more
effective.
4 Roadmap to build a machine learning model

DIVIDE THE DATASET

The entire dataset needs to be divided into


majorly three chunks: training data, validation
data, and testing data

Training data is used to build (fit) the model,


validation data is used to test the model with
known values, and testing data is used to test
the model with unknown values.
5 Roadmap to build a machine learning model

IDENTIFYING ALGORITHM

Based on the dataset, an effective selection of


an algorithm is very important. There are three
types: Regression, Classification, and
Clustering.

Regression and Classification are supervised


learning algorithms (have known labels), while
Clustering is an unsupervised learning
algorithm (no known labels).
6 Roadmap to build a machine learning model

BUILDING THE MODEL

Using the training data, the model is built and


it is further tested with the validation data.

Fine-tuning the model is performed (if need


be) and multiple iterations are performed to
check whether the outcomes are updating or
not.
7 Roadmap to build a machine learning model

EVALUATING THE MODEL

The most important step in building any ML


model is to evaluate it and check for errors.

This step is mandatory to check whether the


algorithm that is applied is effective/accurate
or not and it is done by calculating different
scores, error terms, and other metrics.
8 Roadmap to build a machine learning model

USE TESTING DATA

To further cross-check the model, use testing


data (unknown data values) to generate the
outcomes.

Again evaluate the model and generate


different metrics. If need be fine-tune the
model and reiterate (Go to Step 6).
9 Roadmap to build a machine learning model

BUILD API AND GUI

Once the modelling is completed and the


testing is successful, build application
programming interfaces (APIs).

Connect these APIs with a graphical user


interface (GUI) for the end-user to experience
the machine learning system.
10 Roadmap to build a machine learning model

DEPLOY THE MODEL

Post building the user interface and


connecting all the APIs, the system is ready to
be deployed on a cloud server

A few of the cloud service providers are


Microsoft Azure, Google Cloud Platform, and
Amazon Web Server (AWS). There are others as
well, but these three provide great tools to
manage ML models.
I hope you found

Follow me for more such insightful posts


this roadmap
insightful and useful

Follow Me: Support Me:

Roadmap by Akshay Toshniwal

You might also like