Red Wine Quality Prediction

Brief description of the data set

The red wine dataset is created from the Portuguese "Vinho Verde" wine by gathering information about the wine such as acidity, sugar level, chlorides, etc. to assess its quality. This dataset is highly popular and used for regression analysis for machine learning projects. This dataset includes 12 columns in total. The quality column is the target variable, and the rest of the columns are the related features of the wine measured by chemical tests. The total sample size of this dataset is 1599. All feature columns are float type and the type of the target column is an integer. There are not any missing data in this dataset.

fixed acidity(g/dm^3): It is also known as nonvolatile acid. This type of acid is hard to evaporate. Tartaric acid is one of the fixed acids that play an important role in the color, chemical stability, and taste of the wine.
volatile acidity(g/dm^3): Acetic acid level in the wine is also referred to volatile acidity. It can be created as a by-product of acetic acid fermentation. A high level of acetic acid leads to an unpleasant taste.
citric acid(g/dm^3): Very small quantities can be found in wines. It is produced by fermenting sucrose solutions to boost the acidity of the wine. The taste of freshness and flavor of the wine is determined by the citric acid level.
residual sugar(g/dm^3): Glucose and fructose are the sources of sugar in grapes. Yeast eats this sugar to produce alcohol during fermentation. The level of leftover sugar after this process affects the sweetness of the wine. 35-120 g/L is considered as a medium level of sweetness.
chlorides(g/dm^3): The saltiness of the wine is determined by some organic acids during the production and chlorides play a major role in this level of saltiness. The level of chloride determines the taste of the saltiness of the wine.
free sulfur dioxide(g/dm^3): Sulfur dioxide is critical in making wine which prevents the wine from oxidation and spoilage. The free part is referred to as not bound to other chemicals.
total sulfur dioxide (g/md^3): The total proportion of sulfur includes the free amount plus the bounded part to other chemicals such as aldehydes. The maximum allowable concentration of total sulfur dioxide is 350 mg/L.
density(g/cm^3): It shows the density of the wine which is affected highly by the relative concentration of alcohol and sugar to other ingredients.
pH: The acidity of the wine is determined by the level of pH. The low pH number shows the high acidity while high pH levels are an indicator of the low level of acidity. Typical pH levels for most wines are between 3 and 4.
sulphates(g/dm^3): The natural sulfites are produced during the fermentation process and the added ones are artificially put into the wine to prevent oxidation and unwanted bacteria.
alcohol(% by volume): It shows directly the proportion of alcohol content of the wine. Values shown in this column are the relative percentage of alcohol in the wine.
quality: This is the target column of the dataset and it shows the quality of the wine according to the features measured of the wine.

Main objective: Assessment of the quality of the wine is a very expensive and time-consuming process. Because this process involves human interference, the quality of the wine is highly variable from person to person. For the standardization of quality assessments, it is essential to understand which chemical properties of the wine play a vital role in the evaluation process of the wine. Because of this reason, the main consideration of this project is the interpretation of the features measured in terms of their effects on quality by using Linear Regression.

Data Exploration and Feature Engineering

The first action for data exploration is to check for any missing values in the features and target columns. For this dataset, there are not any missing values for any columns. Detailed information for dataset can be seen in the below table,