問6
頻出度★★★
A社では，顧客の行動や天候，販売店のロケーションなどの多くの項目から成るデータを取得している。これらのデータを分析することによって販売数量の変化を説明することを考える。その際，説明に使用するパラメータをできるだけ少数に絞りたい。このときに用いる分析法として，最も適切なものはどれか。
令和5年度 ITパスポート試験 問6
ストラテジ系企業活動 › 業務分析・データ利活用
This question's explanation and your answers so far.
解説
問 6
Incorrectア
ABC analysis is a method that arranges items such as sales or inventory in descending order of amount or quantity, divides them into ranks A, B, and C according to the cumulative percentage, and identifies the targets to be managed with priority (Pareto analysis). Its purpose is to rank items by importance, and it is not used for the dimensionality reduction that combines many variables into a few.
Incorrectイ
Cluster analysis is a method that gathers data with similar characteristics (customers or samples) and classifies them into several groups (clusters). It groups objects that resemble one another, and does not reduce the number of variables (parameters) used for explanation.
- ウ
正解 The correct answer is 'ウ Principal component analysis'. Principal component analysis is a method that combines many mutually correlated variables (items) into a smaller number of composite variables (principal components) while keeping information loss as small as possible; it is used for what is called dimensionality reduction (reducing the number of variables). In this question, the goal is to explain changes in sales volume from many items such as 'customer behavior, weather, and store location', while the question states that it wants to reduce the parameters used for the explanation to as few as possible. Principal component analysis best matches the aim of consolidating many items into a few principal components for explanation.
Incorrectエ
Correlation analysis is a method that uses indicators such as the correlation coefficient to examine how strong a linear relationship exists between two variables. It only measures the strength of the relationship between variables, and has no function to combine many items into a few composite variables.
ここがポイント
- Principal component analysis = a dimensionality reduction method that consolidates many correlated variables into a few principal components while suppressing information loss. When you see a description like reducing the items (parameters) to a small number, recall principal component analysis.
- Distinguishing the analysis methods: ABC analysis = ranking for priority management; cluster analysis = grouping similar objects; correlation analysis = measuring the strength of the relationship between two variables.
用語