Machine Learning Unit-2 Quiz

Machine Learning 702AI0C012 Unit-2: Data Exploration, Pre-processing and Visualization

• This unit covers data exploration, pre-processing, and visualization techniques for machine learning.
• Topics include missing value treatment, handling categorical data (mapping ordinal features, encoding class labels, one-hot encoding for nominal features), outlier detection and treatment, feature engineering (variable transformation, variable creation, and feature selection), and data visualization (box plots).
• The workflow of machine learning involves:
  • Input data
  • Preparing the model (data exploration and pre-processing)
  • Learning (model selection, training, and tuning)
  • Performance evaluation (testing & validating models)
  • Performance improvement (refining models & using ensembling techniques like bagging and boosting).

Data Types in Machine Learning

• Data is categorized as qualitative (categorical) or quantitative (numerical).
• Categorical data comprises:
  • Nominal data: unordered categories like blood type, nationality, and gender.
  • Ordinal data: ordered categories like customer satisfaction levels, grades, and hardness of metal.
• Numerical data includes:
  • Interval data: data with meaningful intervals, like temperature or date, but no true zero point.
  • Ratio data: data with a true zero point, like height, weight, and salary.

Data Attributes

• Discrete attributes have a finite or countably infinite number of values (e.g., roll number, street number, gender).
• Numeric attributes can take on any value within a range (e.g., student counts, ranks).
• Binary attributes have only two values (e.g., male/female, yes/no).
• Continuous attributes can take on any real number (e.g., length, height, price).

Descriptive Statistics

• Measures of central tendency:
  • Mean: the average of all data values.
  • Median: the middle value when data is ordered.
  • Mode: the most frequently occurring value.
• Measures of dispersion:
  • Range: the difference between the largest and smallest values.
  • Interquartile range (IQR): the difference between the 75th and 25th percentiles.
  • Standard deviation: a measure of how spread out the data is from the mean.
  • Variance

Data Visualization - Box Plots

• Box plots display data through quartiles, median, minimum, maximum, and outliers to show data distribution and skewness.
• Box plots aid in visualizing the five-number summary (minimum, first quartile, median, third quartile, maximum), which shows the center, spread & shape of data.
• They are helpful in identifying outliers, dispersion, mean values, and signs of skewness in given data.

Handling Missing Values

• Three main categories of methods to handle missing values:
  • Skipping: discarding data points or features with missing values.
  • Imputation: replacing missing values with estimated values based on other data points or features.
• Methods for imputation:
  • Global constant
  • Central tendency (mean or median)
  • Most probable value (mode)
• Choosing a method depends on the context and strategy of analysis that would minimize loss or skewing of results when the data is discarded.

Normalization

• Techniques to convert different scales of data to a consistent scale:
  • Min-max normalization: shifts data to range between 0 and 1.
  • Z-score normalization: standardizes data to mean = 0, std dev = 1.

Outlier Treatment

• Three approaches to treat outliers aside from removal:
  • Trimming: reducing the effect of outliers (weights)
  • Mean/Median: replace outlier values using mean/median
  • Log transformation: transform the variable to reduce skewness