ISYE 6501 Mid Term 1 with 100% correct answers

What is modeling? Describing a real life situation mathematically Model Mathematical approaches to solving analytics problems 1. Regression for ex. can be called the model of choice. 2. If more detail (descriptive and numerical), the use of, for ex. Regression, can be called the model. 3. The actual values and output along with 1 and 2 can be referred to as the model. Cross Cutting Concepts Preparing data, measuring the quality of a model's output, missing data, etc.. Classification The process of grouping things based on their similarities Ex. Classifying consumers by who is likely to pay back a loan and who is not Soft Classifier A soft classifier is used when it is not possible to completely separate two categories of data points via the classifying line. the soft classifier gives as good of a separation as possible rather than a hard classifier that separates perfectly. Classifier Rules Must define the best separator. Misclassifying a data point may be more detrimental than misclassifying another. the more costly one bad decision is, the more we want to move away from it. Its okay to push the classifier in one direction so that it is conservative if stakes are higher. Data Point Every Row, a single observation of information Attribute/Feature/Covariate/Predictor Piece of information, column of data Responses/Outcomes Special type of column known as being the answer for each data point Structured Data Can be easily described and stored Unstructured Data Usually in form of text, cannot be easily described or stored Quantitative Data Most numerical data, numbers have a meaning Non Quantitative Data Descriptive data, Categorical data, any data without numerical meaning ex. zip codes Categorical Data numerical data that does not have quantitative meaning Binary Data Subset of categorical data, only 2 values: M/F, On/Off Unrelated data No relationships such as applicants or customer Time series data same data recorded over time, often at different intervals Coefficient a numerical or constant quantity placed before and multiplying the variable in an algebraic expression Support Vector Machine A discriminative classifier formally defined by a separating hyperplane. In other words, given labeled training data (supervised learning), the algorithm outputs an optimal hyperplane which categorizes new data points. SVM margin 1. the line is defined by a set of coefficients a1 through am for each attribute, and an intercept a0. 2. We want to find values of a0 and a1, up to am that classify the points correctly and have the maximum gap or margin between the parallel lines. 3. For each point j we can calculate the error in its calculation. Maximizing the margin involves minimizing the sum over all factors of aj squared.