-
29 October 2007
Appending a String Value Conditionally
-
Dear All,
I am trying to derive a new field as follows :
1)..If a field called "text" has String value " the sun is shining"
2)...and a field called "score" has value 0.9
if score >= 1.0 then deriveField=contents of field "text" + the String "_GREATER"
if score <1.0 then deriveField=contents of field "text" + the String "_LESS"
So in the above example the derived field will have the value
"the sun is shining_LESS"
I tried doing that with a conditional derive node but i haven't made it...any other ideas?
Many Thanks
Harry
-
14 October 2007
I want Apriori algorithm soucre code i want
-
hi firends,
i need apriori algorithm along with the source code in java with all necessary details regarding implementation.thanks in advance.
-
04 October 2007
What is the fastest data input method?
-
Most of the time I use text files for input and output. But actually, when the stream becomes long I would prefer to store all the data in database using ODBC. But this involves reading and writing back to the database many times during stream execution (with multiple streams after each other).
I tested it with a dataset of 100,000 records, each 10 fields, and reading as well as writing is faster when I use simple text files instead of a database. I just test it with the slow MS Access though.
Is this just a fact, that Clementine reads and writes faster from text files or is there any way to speed up the input and output process of a database?
Would the result be much different when using a more powerful database, like Oracle?
Would be happy to read about your thoughts on this.
Ken Aston
-
04 October 2007
fyi - Conference: Effective Data Analysis to Drive Marketing Strategies and Maximise Customer Value
-
There is a Marketing Analytics conference in which I am the opening speaker (god help me...). It is in Singapore during 19-20st November 2007 (http://www.analytics2007.com). I am also giving an associated three hour workshop on 21st November.
I was requested to speak on a fairly generic topic (see below). I will try to cover the major data mining issues many large companies face. I will briefly mention Clementine and Teradata as the two main systems we use to conduct our data mining and advanced analytics. My presentation is 45 mins (including Q&A) so I'll not have much time to cover a lot of material.
My presentation will have a business focus, and involve case studies within the telecommunication industry (in which I now work). Since I am a hands-on technical person there will be some details about the sophistication of our analysis, and practically how we use data to improvement the relationship with our customers.
If you are attending and happen to bump into me, feel free to say hello.
Cheers
Tim
---------------
Presentation (Monday 19th Nov 9am)
IMPORTANCE OF ANALYTICS IN CRM INFRASTRUCTURE
Optus SingTel has stored and analysed data from a Teradata Integrated Data Warehouse (IDW) since 2002. Prior to this, data was only available from legacy billing platforms and the raw telephony switches.
Optus has a Customer Insights analysis team that uses the data warehouse to profile and predict customer behaviour. This analysis often involves processing large amounts of telephony usage data (70 million rows per day) in complex ways.
Tim will be sharing with you the benefits of analytics in:
- Study the importance of marketing analytics in CRM infrastructure development
- Discussion on how analytics can help in marketing a product successfully to its end-user.
- Establish how analytics help in business positioning by being proactive to the needs of customers
WorkShop (Wednesday 21st Nov 9am)
How to Analytically Conduct Data Mining in Marketing Data Warehouse.- it'll be an introduction to datamining with case studies from a teleco and practical examples of how you can change business processes and improve marketing results by using data driven marketing campaigns.
- it'll be an introduction to datamining with case studies from a teleco and practical examples of how you can change business processes and improve marketing results by using data driven marketing campaigns.
-
04 October 2007
Thoughts on my results: how neural networks actually work?
-
I've been making some churn neural networks models, and I've arrived to some conclusions, but this is something I can't understand very well.
The category T I'm trying to predict is very small, so I use a small sample of F for training. I leave default options. So I arrived to two "best models". The model A is the one with the best gain chart considering only the best scored records, and the model B is the one with the best gain chart overall. Model A was trained with 27% records of the T category, and model B was trained with 46% records of the T category. So, is there someway to explain what is happening here or it's just a random better result based on random neural network adjustments (not done by me, I mean algorithm values)?
Thanks!
-
02 October 2007
Balance Node settings question
-
Hello All,
I am new to a Clementine scripting and I could not find any examples on task similar to mine.
So, if anyone knows how to send dynamic parameters into balance node please help me. Thanks!
What I need to do is to change multiplicative parameter based on condition, like that :
set :balancenode.directives=[ {1 "ncounts = 1 " } {2 "ncounts = 2 " } {3 "ncounts = 3 " } \
{4 "ncounts = 4 " } {5 "ncounts = 5 " } {6 "ncounts = 6 " } {7 "ncounts = 7 " } {8 "ncounts = 8 " }]
execute :balancenode
but I do not want to type every single parameter, I am looking for a way to do something like this:
set :balancenode.directives=[{ncounts "'ncounts =' >< ncounts " }] ( syntax is not right)
Thanks,
natalie
-
01 October 2007
Popularity of Neural Networks
-
Neural networks were very fashionable in years past, but I wonder how many people still use them instead of alternative modeling methods (regression, tree-induction, SVM, etc.)? Does anyone here use neural networks? What for?
-
01 October 2007
Intrusion Detection algorithm using NN/Decision Tree/Fuzzy based KDD Cup 99 dataset
-
Hai all.
Has any one of you here have done this project? Attack classifier using artificial intelligent technique (NN/fuzzy/DT/GP...etc) based KDD Cup 99 dataset form MIT Lab.
If anyone of you has been done this classifications, can i have your source code in (c/vb/jave/mathlab..etc)?
plz...
pudin,
KuaLa LumPur.