TEL: 044-952-0102


Getting more from your data with predictive analytics

発行 Ovum (TMT Intelligence, Informa) 商品コード 134054
出版日 ページ情報 英文 21 Pages
納期: 即日から翌営業日
Back to Top
予測分析によるデータの活用 Getting more from your data with predictive analytics
出版日: 2010年07月28日 ページ情報: 英文 21 Pages





  • サマリー
  • 影響
  • Ovumの見解
  • 主なメッセージ
  • 予測分析の事業価値
  • 事業のより深いレベルでの洞察をもたらす予測分析
  • 販売、マーケティングにも拡張する予測分析
  • 事業と技術をミックスさせる動向が予測分析の導入を活性化させる
  • 予測分析が従来のBIと異なる点
  • 予測分析とは
  • 様々な規範を組み合わせた予測分析
  • アプリケーション
  • テクニック
  • アルゴリズム
  • アルゴリズム開発に多大な投資を行うベンダー
  • 監視下および非監視下学習
  • 監視下学習
  • 非監視下学習
  • 現実には予測分析アプリケーションにおいてはハイブリッドなアプローチが用いられる
  • 正しいテクニックの選択が成功の必須要件
  • 継続性とインタラクティブ性を持つ予測分析
  • データ選定
  • データ変換
  • データ探査
  • モデリング
  • 導入
  • モデル管理
  • 技術の許可
  • 全プロセス・ライフサイクルのサポート
  • データの統合と品質
  • データ管理
  • 予測分析のツール
  • 高パフォーマンス構造と加工モデルが拡張性と実績を改善
  • 大規模な平行加工
  • MapReduceとHadoop
  • インデータベース分析
  • 列指向データベース
  • インメモリー
  • クラウド・コンピューティング
  • 提言
  • 企業への提言
  • トップによるコミットメントを必要とする予測分析
  • 予測分析は企業のあらゆる病に対する答えではない
  • データの準備は予測分析における秘密
  • 他のデータ管理構造や加工モデルも考慮を
  • 予測分析モデリング技術の調達方法の検討を
  • 常にエンドユーザーを念頭に
  • ベンダーへの提言
  • ビジネスユーザー向けユーザビリティの改善
  • パッケージ化および産業別の予測分析
  • 付属資料
  • 関連報告書
  • 調査手法
Product Code: 052585


Organizations realize that to compete in an increasingly global and regulated business environment they must find new ways to use and analyze the vast amount of data they have been collecting for so long but not exploited to its fullest effect. Predictive analytics can help improve data analytics over conventional business intelligence (BI) analysis by bringing a new level of intelligence and foresight to the process of turning data into information for competitive advantage.

Research shows that high-performing companies are those that effectively use predictive analytics. It is no coincidence that these companies have experienced significantly higher profit margins and revenue growth and acquire and retain more customers compared with their industry peers. This technology is already proving itself in many industries by helping organizations understand customer behavior, identify unexpected opportunities and threats, and anticipate operational business problems, all before they happen.

Predictive analytics, however, is an exciting but complex technology. It applies a variety of complex algorithms, statistical models, and mathematics to large volumes of data to discover hidden patterns or relationships within that data. Likewise, the data preparation and data modeling aspects of predictive analytics require highly skilled and experienced resources; these are the business analysts and statisticians who need to have a deep understanding of the business and know how to prepare the data, use statistical tools, and interpret the results. These skills are also in short supply, and this makes them expensive.

However, there are possible solutions to these issues. Advances in computing power and processing models such as MapReduce and in-memory and in-database analytics are increasingly being leveraged for computational-intensive tasks such as predictive analytics to provide performance and scale to complex data analysis. Similarly, the cloud and commercial open-source languages such as “R” are helping lower the cost and complexity of predictive analytics.

Predictive analytics brings huge potential to organizations, providing them with greater intelligence and foresight. However, successful implementation relies on organizations realizing the heavy human element to predictive analytics that requires a commitment and desire to invest in skills and use information in new, different, and interesting ways.

Table of Contents

Ovum view
Key messages
Predictive analytics brings a higher degree of insight to your business
Predictive analytics extends beyond sales and marketing
A mix of business and technology trends is driving adoption of predictive analytics
Predictive analytics differs from traditional BI
Predictive analytics combines a mix of disciplines
Vendors invest significant amounts in algorithm development
Supervised and unsupervised learning
Supervised learning
Unsupervised learning
In reality, a hybrid approach is used within predictive analytics applications
Choosing the right technique is vital to success
Predictive analytics is continuous and iterative
Data selection
Data transformation
Data exploration
Model management
Supporting the full process lifecycle
Data integration and quality
Data management
Predictive analytics tools
High-performance architectures and processing models are improving scalability and performance
Massively parallel processing
MapReduce and Hadoop
In-database analytics
Columnar databases
Cloud computing
Recommendations for enterprises
Predictive analytics requires commitment from the top
Predictive analytics is not an answer to all a company’s ills
Data preparation is the dirty secret of predictive analytics
Consider alternative data warehousing architectures and processing models
Consider how you will source predictive analytics modeling skills
Keep end users in mind
Recommendations for vendors
Improve usability for business users
Package and verticalize predictive analytics offerings
Further reading

List of Tables
Table 1: Predictive analytics usage scenarios by function
Table 2: The scope of conventional BI and predictive analytics tools

Back to Top