There are two direction of consideration to complete the matrix with many missing values: statistical models: infer mis

Pre-Questions What? The framework of linear matrix low rank construction Problem? To approximate a matrix with low-rank

I took the course Big Data in the autumen semester 2018, opened by Dr. Ghislain Fourny, Department Information ETH Zuric

Document stores don’t like joins. How can we rebuild the stack with XML/JSON? forced the trees into a table. (Schema-ba

Graph Databases don’t like shards. Issues if using RDMS: the joins are really expensive, not efficient for relations. We

The road to analytics OLTP: consistent and reliable record-keeping. look at detailed individual records (small part of

Use case Write-intensive: highly normalized: avoid update anomalies! Read-intensive: highly denormalized: avoid joins!

XML NavigationOperator slash: / descendant axis : // attribute axis: @attr-name atomization: data(...) filter: [], e.g.@

JobTrackerResponsibilities Resource Management (RM) Scheduling: like ten tasks per machine, who is in charge of reducing

ProcessInput data(key-value pairs) -> split -> Map -> shuffle -> Intermediate data(key-value pairs) ->

SparkIt is for full-DAG query processing. Its first-class citizen: RDD. 8 nodes, 16 cores per node and 128 GB of memory

MeasurementsPrefixes: mili: $m$, 0.001,3, micro: $\mu$, 0.000 001, 6, nano: $n$, 9, pico: $p$, 12, femto: $f$, 15,