1. P K Mallik
New Technologies
Principles, Types, Views, Components, Design
2. P K Mallik
Technology Trends
• Big Data
• Social networks
• Publicly available information
• Large amounts of unstructured data
• Mobile Apps
• Depth rather than coverage
• Internet of Things
3. P K Mallik
Big Data
• Structured to Unstructured
• Centralised to Distributed
• Signal to Noise
• Data Virtualisation
4. P K Mallik
Classification Algorithms
• Define Key attributes on which to
classify
• Create Training Set from earlier
classifications (manual or legacy)
• Run on new data
• Compare attributes with training set
• Make corrections
• Re-Run
• Classify results based on probability
• Send results for analysis
• Customer profiling
• Basket Analysis
• New location selection
• Product aggregation
Retail
• Census Analysis
• Categorization for
commercial offering
Government
• Preventive Medication
• Demographics
Healthcare
• Product
RecommendationInsurance
Classify
Run on
new data
Define
Training
Set
Identify
Key
attributes
Analyse
5. P K Mallik
Segmentation
• Get data points
• Private (e.g. loyalty program)
• Public (government and social media)
• Perform k-Means clustering
• On defined attributes
• 2 to n dimensions
• Interpret clusters
• Buying behaviour
• Perform analysis
• Offers
• Campaigns
• TargetingRetail
• Policy formulation for
benefitsGovernment
• Identifying travel
behaviour for different
segments
• Prospect Demographics
Travel
• Price ComparisonsInsurance
Interpret
clusters
Perform K-
means
clustering
Get data
points
Analyse
6. P K Mallik
Sentiment Analysis
• Create List
• Polarity (+ve or –ve)
• Emotion (joy, anger, sadness etc.)
• Collect Text
• Apply keyword based classification
algorithm
• Create word clouds
• Analyse
• Plot graphs
• Create profiles
• Identify trends
• Customer feedbackRetail
• Impact of policies
• Public view of
proposals
Government
• Trends
• Impact of events
Travel
• Customer Feedback
• Impact of events
Insurance
Create
word
clouds
Apply
classificat
ion
algorithm
Collect
Text
Create
list Analyse
7. P K Mallik
Competitive Analysis
• Competition is identified manually
• Identify the Product SKU’s for
which you want the analysis
• Crawl the web to collect
information from competitive sites
• Tabulate price or ratings
information
• Send data for analysis (trends,
market share etc.)
• Price AnalyticsRetail
• Cross Country
BenchmarkingGovernment
• Price AnalyticsTravel
Tabulate
Informat
ion
Crawl
the Web
Define
Product
Identify
Compe
tition
Analyse
8. P K Mallik
Text Analysis
• Get dictionary
• Key value pairing
• Read Text
• Any document
• Website
• Plain text
• Extract Concepts
• Apply rules
• Match words
• Convert to structured data
• Send for analysis
• Product Trend AnalysisRetail
• Medical UnderwritingInsurance
• Travel Trend AnalysisTravel
Get
structure
d data
Extract
concepts
Read text
Get
dictionar
y
Analyse
9. P K Mallik
Data Enrichment
• Identify Data Sources
• Get data for targeted
group
• Augment existing
database
• Customer single view
• Customised services
and offers
• Deep dive analysis
Retail
• Enhance census data
• Impact of policies
• Affected segments for
proposed policies
Government
• Auto complete forms
• Targeted updates
Insurance
Augment
existing
database
Get data
for
targeted
group
Identify
Data
Sources
Utilise
10. P K Mallik
User Experience
• User Experience
• Social Media Vs Portals
• Push instead of Pull
• Engagement Vs Throughput
• Vanishing Keyboard
11. P K Mallik
Social Vs Portals
• Interaction to Collaboration
• Transaction to Knowledge
• Status to Events
• User Single View to 360 degree view
• User desktop
• Community driven App space
12. P K Mallik
Push instead of pull
• AI driven notifications
• Most suitable device
13. P K Mallik
Engagement Vs Throughput
• Stickiness to Gamification
• Peripheral or secondary use
• Reveal more information
• Usage analysis
• Cross Sell/Upsell to Lifetime Value
• From selling process to participating in buying journey
14. P K Mallik
Vanishing Keyboards
• Touch increasingly becoming default
• Voice gaining popularity
• Thought control
15. P K Mallik
Internet of Things
• Connected devices
• Wearables
• Cars
• Homes
• Cities
• Industrials
• Reduction of human intervention
• Fog Computing Architecture
• Most data will be noisy and latency sensitive
• Too expensive to be carried back to the central servers or cloud