Training objectives:
Curriculum theory and experience from the induction of primary data mining projects often encounter problems, analysis and summary, given targeted solutions, will reproduce the classic case of course these questions, explain by example, and corresponds to the student the actual work problems, so that students will be able to combine the experience and teach them their own problems, effectively inspired ideas, stimulate interest, new methods and new ideas to solve the problem needs to provide.
Students harvest summed up the three points:
1: Understanding the actual case in conjunction with said various variants of the basic algorithm, open ideas
2: a training to deepen a channel depth understanding of a breakdown field after can receive training knowledge point corresponding industry's latest update.
3: Add "data mining" professional circle, students can have a chance to join professional social circles, and the trade exchange, learn from each other.
Principles and applications of generalized linear classification of
content: logistic Regression and computational advertising
. Examples of binary classification principles n2 mathematical logic regression of
traditional advertising, ad, advertising, real-time trading platform for
data and pre-dimension extraction
applications in computing LR ads
evaluation of the effect of LR model tuning
compromise LR model theory and engineering of
LR and his little friends: generalized linear model n9 new situation, based on mobile advertising end real-time trading platform, LR various variants of
the decision tree, clustering and anomaly detection
content: decision trees and outlier detection
principle decision tree
various decision tree algorithm generates
decision trees in anomaly detection of
tree pruning
application tips detection of other abnormalities
clustering algorithm the principle of
common problems clustering algorithm: initial point selection,
clustering algorithms and decision trees in anomaly detection combination of computing advertising, talk about two small examples outlier detection
giant graph mining
complex network description
of the current giant map scenarios
random FIG, FIG natural,
commonly FIG computational framework (Google is bagel, graphlab the Graphx)
algorithm giant FIG. Now the principle of
common graph-based algorithm to achieve
a random walk, pageRank based implementation map
svd svd in the introduction and implementation of FIG mining and recommendation system view of the frame (qzone advertising system wide-point, the contents twitter recommended)
Telephone counseling: 010-62883247,62884854 br /> Mail Advisory: [email protected]
Zhongkexin soft Senior Technical Service Training Address: Yangfangdian Road, Haidian District, Beijing Oriental Plaza, No. 18, Block N 520/521 shine.