Supervised Machine learning

What is Machine Learning?

Machine learning is a core area under artificial intelligence  Machine learning (ML)  allow the computer to learn the data and predict without being programmed by human intervention, hare Machine is referred to model and learning refer to input dataset.

Although Machine learning technology is not new, it is now growing fresh momentum as there are so many things to know about ML. Today, machine learning is different from what it used to be in the past.

In Past days where programmers code a machine how to solve a problem. Now we are in the era of machine learning where machines are automatically trying to solve problems, by their own, by identifying the trends and patterns in each data set and to predict future problems and their solutions.

Supervised Machine Learning

 Why Machine Learning?

To better identify the uses of machine learning, consider some of the instance where machine learning is functional: the self-driving car, stock price, email spam cyber fraud finding, online recommendation engines like friend submissions on Facebook, Netflix suggest the movies and shows you might like, Amazon and other online store Advertisement which are coming on most of the website are all examples of applied machine learning.

Nowadays, corporations can make ML algorithms to develop analytical models, trends, and patterns with minimal or no human intervention.

The factors of growing interest in ML are:-

  1. continuously growing volumes of huge dataset Big Data ( Volume, Velocity, Variety).90% of the data that we have today has been generated in last 2 years
  2. Real Time and Customized Decisioning- Algorithm driven positioning of products and customer targeting.
  3. powerful and affordable computational processing
  4. affordable data storage options.
  5. Multi-Modality and Heterogeneity – Data comes from different platforms and in all shapes and forms such as videos, text, images, social interactions, comments and so on.

Machine Learning Algorithm

Generally, there are 3 types of learning algorithm:

  1. Supervised Machine Learning Algorithms
  2. Unsupervised Machine Learning Algorithms
  3. Reinforcement Machine Learning Algorithms

In this article, we are going to discuss the longer term of Machine Learning and understand why we should learn Machine Learning. We will also discuss Supervised Machine Learning Algorithms and the limitations and advantages. Conjunction with this we will also look at real-life Machine Learning applications.

Supervised Machine learning as the name specifies a presence of supervisor as teacher teach the machine how to learn. These algorithms are trained using labeled dataset, Supervised machine learning algorithms dataset are pre-labeled and define the relationship between data or feathers, patterns, all data are arranged under the unique labeled name. like name, phone number, ID, etc.

Supervised Machine Learning, input data is so-called training data and has a predefined label such as spam/not-spam or a stock price at a time.

Supervised Machine Learning

Since we have provided labeled data during training set model is able to make predictions about future outcomes based on historical data  This algorithm is commonly used in areas where historical data is used to predict that are likely to occur in the future. An algorithm will receive a set of input data then it compares the actual output with the accurate output and flag an error.

The common example of handwriting recognition is a supervised learning algorithm. We display the computer several numbers of images of handwritten with the correct labels for those digits, and the computer recognizes the patterns and relate images with their labels.

In simple word supervised learning is a type of learning where we educate the machine with data which we already labeled, it means some data is already labeled with the known result. After that, the machine is projected with the new dataset that’s how supervised learning algorithm (machine) examines the training data and produces a correct output from labeled data.

For example, Suppose you have a fruits basket with different varieties of fruits, the first step to train the model by showing different fruits one by one.

Machine Learning

NO. SIZE COLOR SHAPE FRUIT NAME
1 Big Red Depression at the top and Rounded shape Apple
2 Small Red Red Heart-shaped Strawberry
3 Big Green Yellow Long curving cylinder Banana
4 Small Green Bunch shape Cylindrical, Round to oval Grape

 

  • If the color of the fruit is Orange and shape is rounded then it will be labeled as –Orange.
  • If the color of the fruit is Red and Heart-shaped then it will be labeled as –Strawberry.

Now after training the data, you are given a new fruit say Orange from the basket and asked to classify it.

Supervised Machine Learning

 

Sine the machine has already trained and understand the variation in data, if you show any new fruit it will first group the fruit based on color and then shape and predict the new fruits would be Orange.

So this is how the machine learns the things from training data(fruits basket) and then apply the knowledge to test data(new fruit).

So Before we go in details just have a feel the power of Machine Learning Drawing.

  1. teachablemachine.withgoogle.com
  2. quickdraw.withgoogle.com
  3. autodraw.com
  4. sketch_rnn_demo

Supervised Machine Learning Examples 

Regression:- The Regression algorithm is normally used to predict the numeric data in its place of labels. It can also classify the distribution movement depending on the available historical data the algorithm returns a numerical value for each value, regression is useful for predicting outputs that are continuous.  such as stock price, predict housing prices, spam/not-spam, and revenue generated by the campaign.

Algorithms for Regressions?

      • Linear Regression
      • Logistic Regression
      • Polynomial Regression
      • Random forest
      • Stepwise Regression
      • Support Vector Regression (SVR)
      • Ridge Regression
      • Lasso Regression

 

Classification: -The main goal of classification is to predict the target class (Yes/ No). in which the algorithm effort to label each data by selecting between two or more, unlike classes. In a binary classification model selecting between two classes such as finding whether an email is spam or not, or whether or not person will loan defaulter, predict whether the student will pass or fail, predict whether the customer will buy the new product or not, the patient has cancer or not, image contains a dog or not have only two possible outcomes (Yes/ No) this is called binary classification. Classification algorithms are applied whenever the desired output is separate by the label. Many use cases, image and audio categorization, customer segmentation, and text analysis for mining customer sentiment.

Algorithms for classification

      • Logistic Regression
      • Decision Trees
      • K Nearest Neighbors
      • Naive Bayes
      • Linear SVC (Support vector Classifier)

Comparison Chart

BASIS FOR COMPARISON CLASSIFICATION REGRESSION
Basic The model or algorithm when the mapping of data is done into predefined classes(labeled). The model in which the mapping of data is done into values.
Includes prediction of Discrete data values Continuous data values
Algorithms logistic regression, SVC, Decision tree, etc. Linear regression, Regression tree (Random forest), etc.
Nature of the predicted data Unordered Ordered
Mode of calculation Measuring accuracy Measurement of root mean square error

Supervised Machine Learning

Supervised Machine Learning Applications

Supervised machine learning is one of the most powerful model, which give more accurate and faster prediction compare to humans. Businesses industries use it to solve problems like:

 

      1. WASP predicts the chances of winning team based on various features like player-vs-player, pitch condition, toss, head-to-head and team and player past record.
      2. Train OCR model system for handwriting and once it fully trained it is capable to convert handwriting images into text with some level of accuracy.
      3. Speech to text reorganization mechanized system in your mobile trains human voice and then starts working based on this training convert speech voice input to text output.
      4. Based on some prior knowledge weather application predict based on higher temperature means sunny and higher humidity means cloudy.
      5. Your email is well-filtered spam mail based on its past training info, So if any new email comes it will automatically be categorized into Spam mail, normal mail or social media mail, Promotional mail.
      6. Control Successpect things where if you use your Bank credit card in one state or country and if you try to use in another country you will receive a message to show your identity or if you always use one device (Mobile, laptop) to check your emails and by chance if you log in your account from another device you will receive an alert message.

Wrapping up

Hope you enjoy !!!! this article so in the last conclusion is Supervised Machine Learning in which someone supervised or teach the machine/ model with already labeled data and based on model understanding, it will predict the future data.

You May Also Like