在企業(yè)里面從事大數(shù)據(jù)相關(guān)的工作到底需要掌握哪些知識(shí)呢?我認(rèn)為需要從兩個(gè)角度來(lái)看:一個(gè)是技術(shù);一個(gè)是業(yè)務(wù)。技術(shù)上主要涉及到概率和數(shù)理統(tǒng)計(jì),計(jì)算機(jī)系統(tǒng)、算法和編程等;而業(yè)務(wù)的角度呢則是因公司業(yè)務(wù)的不同而異。對(duì)于從事大數(shù)據(jù)的工程人員來(lái)說(shuō),需要學(xué)
在企業(yè)里面從事大數(shù)據(jù)相關(guān)的工作到底需要掌握哪些知識(shí)呢?我認(rèn)為需要從兩個(gè)角度來(lái)看:一個(gè)是技術(shù);一個(gè)是業(yè)務(wù)。技術(shù)上主要涉及到概率和數(shù)理統(tǒng)計(jì),計(jì)算機(jī)系統(tǒng)、算法和編程等;而業(yè)務(wù)的角度呢則是因公司業(yè)務(wù)的不同而異。對(duì)于從事大數(shù)據(jù)的工程人員來(lái)說(shuō),需要學(xué)會(huì)使用數(shù)據(jù)挖掘方法在計(jì)算機(jī)系統(tǒng)和編程工具的幫助下解決實(shí)際的問(wèn)題,這樣才能夠在海量數(shù)據(jù)中挖掘出業(yè)務(wù)增長(zhǎng)的助推劑,才能在激烈的市場(chǎng)競(jìng)爭(zhēng)中為企業(yè)創(chuàng)造更多的價(jià)值。
因?yàn)闃I(yè)務(wù)會(huì)因公司的不同而不同,但是技術(shù)點(diǎn)是想通的。我在這里簡(jiǎn)單總結(jié)了一下大數(shù)據(jù)相關(guān)工程人員需要掌握的技術(shù)相關(guān)知識(shí)點(diǎn)。主要涉及到數(shù)據(jù)庫(kù)、數(shù)據(jù)倉(cāng)庫(kù)、編程、分布式系統(tǒng)、Hadoop生態(tài)系統(tǒng)相關(guān)、數(shù)據(jù)挖掘和機(jī)器學(xué)習(xí)相關(guān)的基礎(chǔ)知識(shí)點(diǎn)。當(dāng)然我這里列出來(lái)的應(yīng)該是一個(gè)team的人員匯集在一起所具備的,每個(gè)人會(huì)因在團(tuán)隊(duì)中的角色不同而有所側(cè)重。在此剖磚引玉,歡迎大家發(fā)表意見(jiàn)。
Topic | Content | Key points | Reference | |
DB/OLTP & DW/OLAP | Database/OLTP basic | The relational model, SQL, index/secondary index, inner join/left join/right join/full join, transaction/ACID | Ramakrishnan, Raghu, and Johannes Gehrke. Database Management Systems. | |
Database internal & implementation | Architecture, memory management, storage/B+ tree, query parse /optimization/execution, hash join/sort-merge join | |||
Distributed and parallel database | Sharding, database proxy | |||
Data warehouse/OLAP | Materialized views, ETL, column-oriented storage, reporting, BI tools | |||
Basic programming | Programming language | Java, Python (Pandas/NumPy/SciPy/scikit-learn), SQL, Functional programming, R/SAS/SPSS | Wes McKinney. Python for Data Analysis: Agile Tools for Real World Data. | |
OS | Linux | |||
DB & DW system | MySQL/ Hive/Impala | |||
Text format and process | JSON/XML, regex | |||
Tool | Git/SVN, Maven | |||
Distributed system & Hadoop ecosystem & NoSQL | Distributed system principal theory | CAP theorem, RPC (Protocol Buffer/Thrift/Avro), Zookeeper, Metadata management (HCatalog) | ||
Distributed storage & computing framework & resource management | Hadoop/HDFS/MapReduce/YARN | Tom White. Hadoop : The Definitive Guide.
Donald Miner, Adam Shook. MapReduce Design Patterns : Building Effective Algorithm and Analytics for Hadoop and Other Systems. |
||
SQL on Hadoop | Data (log) acquisition/integration/fusion, normalization, feature extraction | Sqoop, Flume/Scribe/Chukwa,SerDe | Edward Capriolo, Dean Wampler, Jason Rutherglen. Programming Hive. | |
Query & In-database analytics | Hive, Impala, UDF/UDAF | |||
Large scale data mining & machine learning framework | Spark/MLbase, MR/Mahout | |||
Streaming process | Storm | |||
NoSQL | HBase/Cassandra (column oriented database) | Lars George. HBase: The Definitive Guide. | ||
Mongodb (Document database) | ||||
Neo4j (graph database) | ||||
Redis (cache) | ||||
Data mining & Machine learning | DM & ML basic | Numerical/Categorical variable, training/test data, over fitting, bias/variance, precision/recall, tagging | ||
Statistic | Data exploration (mean, median/range/standard deviation/variance/histogram), Continues distributions (Normal/ Poisson/Gaussian), covariance, correlation coefficient, distance and similarity computing, Bayes theorem, Monte Carlo Method, Hypothesis testing | |||
Supervised learning | Classifier, boosting, prediction, regression analysis |
Han, Jiawei,Micheline Kamber, and Jian Pei.?Data mining: concepts and techniques.
|
||
Unsupervised learning | Cluster, deep learning | |||
Collaborative filtering |
Item based CF, user based CF
|
|||
Algorithm | Classifier | Decision trees, KNN (K-Nearest neighbor), SVM (support vector machines), SVD (Singular Value Decomposition), na?ve Bayes classifiers, neural networks, | ||
Regression | Linear regression, logistic regression, ranking, perception | |||
Cluster | Hierarchical cluster, K-means cluster, Spectral Cluster | |||
Dimensionality reduction | PCA (Principal Component Analysis), LDA (Linear discriminant Analysis), MDS (Multidimensional scaling) | |||
Text mining & Information retrieval | Corpus, term document matrix, term frequency & weight, association rules, market based analysis, vocabulary mapping, sentiment analysis, tagging, PageRank, VSM (Vector Space Model), inverted index | Jimmy Lin and Chris Dyer. Data-Intensive Text Processing with MapReduce. |
原文地址:大數(shù)據(jù)工程人員知識(shí)圖譜, 感謝原作者分享。
聲明:本網(wǎng)頁(yè)內(nèi)容旨在傳播知識(shí),若有侵權(quán)等問(wèn)題請(qǐng)及時(shí)與本網(wǎng)聯(lián)系,我們將在第一時(shí)間刪除處理。TEL:177 7030 7066 E-MAIL:11247931@qq.com