Wine Data Analysis
Executive Summary
Project Background & Objectives
Rosette Wine Company Ltd. owns one winery while main business is to trade the wines produced from west coast United States. Rosette is expecting to develop a model to predict the price of wines so that they are able to build some inventory when certain wine is offered undervalue like promotion or market new entry.
We, Analytica Inc., are invited by Rosette Wine Co. Ltd. to work out that model based on the database generated by Rosette Wine together with Wine Spectator for three best-sell varietals: Chardonnay, Merlot and Cabernet from 18 most popular wineries in west coast US regions.
Methodology
This dataset contains 10 variables and 890 observations. Through regression of this dataset, an equation is created to quantify the correlation between different variables and wine value and therefore make the preudiction of wine value possible.
712 samples are used in the multiple regression models to generate the formula and 178 holdout samples are used to simulate the accuracy of the prediction resulting from the formula, which turns out to be good.
Key Business Findings
Key finding 1: The formula can be used to help Rosette Wine Co., Ltd. to search for wines whose value is under estimated or in good promotion deal.
Key finding 2: The formula can assist Rosette Wine to improve their pricing mechanism on the wine produced in their own winery with maximized profitability.
Dataset and Variables Definition
The data we use is created in June 2009 with information generated by Rosette Wine together with Wine Spectator. The dataset consists of data for three varietals, namely Chardonnay, Merlot and Cabernet; Four regions from west coast United States, namely Napa Valley, Sonoma Valley, Bay Area and Oregon and then 4-5 vineyards from each region based on the following criteria: they each needed to produce all of the three varietals with corresponding
Project Background & Objectives
Rosette Wine Company Ltd. owns one winery while main business is to trade the wines produced from west coast United States. Rosette is expecting to develop a model to predict the price of wines so that they are able to build some inventory when certain wine is offered undervalue like promotion or market new entry.
We, Analytica Inc., are invited by Rosette Wine Co. Ltd. to work out that model based on the database generated by Rosette Wine together with Wine Spectator for three best-sell varietals: Chardonnay, Merlot and Cabernet from 18 most popular wineries in west coast US regions.
Methodology
This dataset contains 10 variables and 890 observations. Through regression of this dataset, an equation is created to quantify the correlation between different variables and wine value and therefore make the preudiction of wine value possible.
712 samples are used in the multiple regression models to generate the formula and 178 holdout samples are used to simulate the accuracy of the prediction resulting from the formula, which turns out to be good.
Key Business Findings
Key finding 1: The formula can be used to help Rosette Wine Co., Ltd. to search for wines whose value is under estimated or in good promotion deal.
Key finding 2: The formula can assist Rosette Wine to improve their pricing mechanism on the wine produced in their own winery with maximized profitability.
Dataset and Variables Definition
The data we use is created in June 2009 with information generated by Rosette Wine together with Wine Spectator. The dataset consists of data for three varietals, namely Chardonnay, Merlot and Cabernet; Four regions from west coast United States, namely Napa Valley, Sonoma Valley, Bay Area and Oregon and then 4-5 vineyards from each region based on the following criteria: they each needed to produce all of the three varietals with corresponding