Escolar Documentos
Profissional Documentos
Cultura Documentos
Mc130201353
Usman Akram
Mc130201353
An Abstraction
A Representation Which of the following option is true?
Usman Akram
Mc130201353
De-Normalization normally
speeds up Data Retrieval
(Page 51)
Data Modification
Development Cycle
Data Replication
For a given data set, to get a global view in un-supervised
learning we use One-way Clustering (Page 271)
Bi-clustering
Pearson correlation
Euclidean distance
Usman Akram
Mc130201353
Load
Transform
Constraint page 161
Extraction
UAT stands for
Usman Akram
Mc130201353
Inverted
Bitmap page 233
Dense
Sparse index
In context of web data ware house. Which is NOT one of way to identify session
Others Mcqs from Midterm + 3, 4 was from Handouts but very easy .
Subjective:
2 Marks Questions
There are four categories of data quality improvement. Write any two. (2
marks)
Answer:
The four categories of Data Quality Improvement
Process
System
Policy & Procedure
Data Design
Write two unsupervised learning? 2 marks
Answer:
One way clustering
Two way clustering
Usman Akram
Mc130201353
3 Marks Questions
As the number of processes increase, the speedup should also increase. Thus
theoretically there should be a linear speedup; however this is not the case in
real. List at least 2 barrier of linear speedup. 3 marks
Answer:
Amdahl Law
Startup
Interference
Skew
Common Dimensions in context with Web data warehouse. 3marks
Answer: ------Name of three DWH development methodologies? 3 marks
Answer:
Usman Akram
Mc130201353
Data Driven
Goal Driven
User Driven
One question was to identify statement is correct or not (was from Midterm) 3
marks
5 Marks Question:
Before sitting down with the business community to gather information, it is
suggested to set you up for a productive session. Write three activities
requirement preplanning phase 5 marks
Answer:
Requirements preplanning: This phase consists of activities like choosing the forum, identifying and
preparing the requirements team and finally selecting, scheduling and preparing the business
representatives.
This query was given SELECT*FROM R WHERE A= 5 and we have to tell which
Technique is appropriate from dense, sparse, B-tree and has indexing. 5 marks
Answer: Hash Indexing is appropriate for the given query, because hash indexing is good for matching
queries.
According to Amdahls Law prove that the speedup does not remain same if
the fraction of the problem and number of processors are doubled. Please note
that 0 overhead and perfect parallelism is used. Use following examples 5
marks
a)
Fraction of the problem that must be computed sequentially is 5% and
number of processors is 100.
b)
Fraction of the problem that must be computed sequentially is 10% and
number of processors is 200.
Ans: REF (Handouts Page # 204,205)
Amdahls law: S 1 / f + ( 1 f ) / N
a)
b)
Hence it is evident that the speedup does not remain same if we double the fraction and number of
processors.
Usman Akram
Mc130201353
Usman Akram
Mc130201353
modeling is an example of supervised learning because type and number of classes are
known in advance
Question # (06): (3 Marks)
Name any three Data warehouse Development methodologies?
Development methodologies Waterfall model Spiral model RAD Model
Structured Methodology Data Driven Goal Driven User Driven
Question # (07): (3 Marks)
List down any three ways of Handling missing data during Data cleansing process
DWH major issues of data cleansing had arisen due to data processing and handling
at four levels by different groups of people i.e.
(i) Hand recordings by the scouts at the field level
(ii) typing hand recordings into data sheets at the DPWQCP office
(iii) photocopying of the scouting sheets by DPWQCPpersonnel and finally
(iv) Data entry or digitization by hired data entry operators.
Question # (08): (3 Marks)
HTTPs secure socket layer (SSL) is used to identify the session on the world wide web
however the are some limitation of this technique, Briefly explain the two limitations?
Offers an opportunity to track a visitor session
Limitations
To track the session, the entire information exchange needs to be in high
overhead SSL
Each host server must have its own unique security certificate.
Visitors are put-off by pop-up certificate boxes.
This offers an opportunity to track a visitor session because it may include a login action
by the visitor and the exchange of encryption keys.
Limitations
The downside to using this method is that to track the session, the entire
information exchange needs to be in high overhead SSL, and the visitor may be
put off by security advisories that can pop up when certain browsers are used.
Each host server must have its own unique security certificate.
Question # (09): (5 Marks)
Briefly explain any two types of precedence constraints that we can use in DTS?
Unconditional: If you want Task 2 to wait until Task 1 completes, regardless of the
outcome, link Task 1 to Task 2 with an unconditional precedence constraint.
On Success: If you want Task 2 to wait until Task 1 has successfully completed, link Task
1 to Task 2 with an
On Success precedence constraint.
On Failure: If you want Task 2 to begin execution only if Task 1 fails to execute
successfully, link Task 1 to Task 2 with an On Failure precedence constraint. If you want
to run an alternative branch of the workflow when an error is encountered, use this
constraint.
Question # (10): (5 Marks)
Usman Akram
Mc130201353
Briefly explain merge/purge problem while applying data cleansing in data warehousing field?
Within the data warehousing field, data cleansing is applied especially when several
databases are merged. Records referring to the same entity are represented in different
formats in the different data sets or are represented erroneously. Thus, duplicate records
will appear in the merged database. The issue is to identify and eliminate these
duplicates.
The problem is known as the merge/purge problem. Instances of this problem appearing
in literature are called record linkage, semantic integration, instance identification, or
object identity problem.
Question # (11): (5 Marks)
Give two reason, why Rapid Application Development (RAD) is more suitable for data
warehousing development as compared to others traditional development methodologies?
Rapid Application Development (RAD) is an iterative model consisting of stages
like scope, analyze, design, construct, test, implement, and review. It is much better
suited
to the development of a data warehouse because of its iterative nature and fast
iterations.
User requirements are sometimes difficult to establish because business analysts are too
close to the existing infra-structure to easily envision the larger empowerment that data
warehousing can offer. Development and delivery of early prototypes will drive future
requirements as business users are given direct access to information and the ability to
manipulate it. Management of expectations requires that the content of the data
warehouse be clearly communicated for each iteration.
CS614 Current FinalTerm Paper 25 August 2016
Q.What may be possible implications if the developing organization never freezes the
requirements throughout the DWH development i.e. it always behaves like an accommodating
person. 5
Write down any two drawbacks if Date is stored in text format rather than using proper date
format like dd-MMM-yy etc. 5
There are different data mining techniques e.g. clustering, description etc. Each of the
following statement corresponds to some data mining technique. For each statement name
the technique the statement corresponds to. 5
a) Assigning customers to predefined customer segments (i.e. good vs. bad)
b) Assigning credit applicants to predefined classes (i.e. low, medium, or high risk)
c) Guessing how much customers will spend during next 6 months
Usman Akram
Mc130201353
d) Building a model and assigning a value from 0 to 1 to each member of the set. Then
classifying the members into categories based on a threshold value.
e) Guessing how much students will score more than 65% grades in midterm.
a) Assigning customers to predefined customer segments (i.e. good vs.
bad) classification
b) Assigning credit applicants to predefined classes (i.e. low, medium, or high risk)
classification
c) Guessing how much customers will spend during next 6 months prediction
d) Building a model and assigning a value from 0 to 1 to each member of the set.
Then
classifying the members into categories based on a threshold value. Estimation
e) Guessing how much students will score more than 65% grades in midterm.
Prediction
There are two justifications for a task to be performed in parallel, either it manipulates
significant amount of data (i.e. size) or it can be solved by divide and conquer (D&C) strategy.
From the given list, provide the justification for each of the task to perform it in parallel. 5
a) Large table scans and joins
Size
Size
D&C
D&C
What are the tasks performed through import / export data wizard to load data?Write any three
3
Import and Export Data Wizard provides the easiest method of loading data.
The wizard creates package which is a collection of tasks
Tasks can be as follows:
Establish connection through source / destination systems
Creates similar table in SQL Server
Extracts data from text files
Apply very limited basic transformations if required
Loads data into SQL Server table
In context of Four Cell Quadrant Technique, which business process (from the diagram below)
will have highest priority? Justify with reason. [Marks 3]
So the process having higher feasibility and impact is given higher priority over the
process having lower feasibility and impact. In example of Figure 33.2, process A has
highest priority while the process D has lowest priority PG# 298
Usman Akram
Mc130201353
Consider the following two statements. Specify that each statement correspond to which
activity of data quality analysis project. 3
a) Identify functional user data quality requirements and establish data quality
metrics.
Define
b) Measure conformance to current business rules and develop exception reports.
Measure
Identify the given statement as correct or incorrect and justify your answer in either case.
Bayesian Modeling is an example of Unsupervised Learning. 3
The problems associated with the extracted data can correspond to non-primary keys. List
down any four problems associated with the non-primary key.5
Non primary key problems
1. Different encoding in different sources.
2. Multiple ways to represent the same information.
3. Sources might contain invalid data.
4. Two fields with different data but same name.
1. Data may be encoded differently in different sources. The domain of a gender field
in some database may be {F, M} or as {Female, Male} or even as {1, 0}.
2. There are often multiple ways to represent the same piece of information. FAST,
National University, FAST NU and Nat. Univ. of Computers can all can be found
in the literature as representing the same institution.
3. Sources might contain invalid data. A point of sale terminal may require that the sales
clerk enters a customers telephone number. If the customer does not wish to give it,
clerks may enter 999-999-9999.
4. Two fields may contain different data but have the same name. There are a couple of
ways in which this can happen. Total Sales probably means fiscal year sales to one part
of an enterprise and calendar year sales to another. The second instance can be much
more dangerous. If an application is used by multiple divisions, it is likely that a field that
is necessary for one business unit is irrelevant to another and may be left blank by the
second unit or, worse, used for otherwise undocumented purposes.
What is Reverse Proxy?2
Reverse Proxy Another type of proxy server, called a reverse proxy, can be placed in
front of our enterprise's Web servers to help
them offload requests for frequently accessed content. This kind of proxy is entirely
within our control and usually presents no
impediment to Web warehouse data collection. It should be able to supply the same kind
of log information as that produced by a
Web server.
Why analytics track is called as the fun part" while designing a data warehouse?2
Usman Akram
Mc130201353
Usman Akram
Mc130201353
Usman Akram
Mc130201353