Welcome back, we already discussing Classification models and techniques for next few lectures. Let's first define what the classification task is, after this video you will be able to define what classification is. Explain whether classification is supervised or unsupervised and describe how binomial classification differs from multinomial classification. Classification is one type of machine learning problems. In the classification problem, the input data is presented to the machine learning model and the task is to predict the target corresponding to the input data. The target is a categorical variable, so the classification task is to predict the category or label of the target given the input data. For example, the classification problem illustrated in this image is to predict the type of weather. The target that the model has to predict is the weather and the possible values for weather in this case is Sunny, Windy, Rainy, or Cloudy. The input data can consist of measurements like temperature, relative humidity, atmospheric pressure, wind speed, wind direction, etc. So, given specific values for temperature, relative humidity, atmospheric pressure, etc., the task for the model is to predict if the weather will be sunny. Windy, rainy, or cloudy for the day, this is what the data set might look like for the weather classification problem. Each row is a sample with input variables temperature, humidity, and pressure and target variable weather. Each row has specific values for the input variables and a corresponding value for the target variable. The classification task is to predict the value of the target variable from the values of the input variables. Since a target is provided, we have labeled data and so classification is a supervised task. Recall that in a supervised task, the target or desired output for each sample is given. Note that the target variable goes by many names such as target, label, output, class variable, category, and class. A classification problem can be binary or multi-class with binary classification the target variable has two possible values, for example yes and no. With multi-class classification the target variable has more than two possible values. For example, the target can be short, medium and tall. Multi-class classification is also referred to multinomial or multi-label classification. Remember though that the target is always a categorical variable in classification. Some examples of binary classification are predicting whether it will rain tomorrow or not, here there are two possible outcomes, yes it will rain tomorrow or no, it will not rain tomorrow. Identifying whether a credit card transaction is legitimate or fraudulent, again, there are only two possible values for the target, legitimate or fraudulent. Some examples of multi-class classification include predicting what type of product that a customer will buy. The possible values for the target variables would be product categories such as kitchen, electronics, clothes, etc. There is more than one category of products, so this is a multi-class classification problem. Another example is categorizing a tweet as having a positive, negative, or neutral sentiment, again, the number of possible values for the target is more than two here. So, this is also a multi-class classification task to summarize in classification, the model has to predict the category corresponding to the input data. Since the target is provided for each sample, classification is a supervised task, the target variable is always categorical in classification.
