|
|
 |
Mine your own data!
In order to understand the benefits of Data Mining, and
immediately appreciate how ALICE d'ISoft can enhance
your business performance, there's no better way than mining your own
data. That's exactly what we're offering today!
- It's free! You'll be able to extract hidden information from
your data AT NO CHARGE.
- It's easy! Get the evaluation version, and send us your data
by email. We'll send you back an ALICE d'ISoft
project file which you can open with the evaluation version. You'll
then have all the features of ALICE d'ISoft available
with your own data! Build as many decision trees as you want, modify
them manually, create reports, SQL queries and make prediction with
the What-If module.
Please note that this program is restricted to one file per company.
Here is the whole process
- If you haven't already done so, get
an evaluation version of ALICE d'ISoft.
- Select one of your most interesting data files. Format it according
to our instructions (see below).
- Write a small description of your data, specifying:
- The nature of the records (customers, patients, products, companies,
etc.)
- The number of records and the number of fields, so that we can
check if we received your whole file successfully.
- The type of each field. ALICE d'ISoft
detects data types automatically, but certain types (such as ordered
symbolic or symbolic coded with numbers) cannot be detected.
- The field(s) you want explained (which can also be of any type),
that is to say, the question you want ALICE
d'ISoft to answer. Make sure you have understood what a Decision
Tree is before selecting your data.
- Send the data file to register@isoft.fr,
as an attached file. The body of the message must contain the data description
as explained in the previous point, and your full signature.
- We will return the project file containing your data. Launch the
ALICE d'ISoft evaluation version and open this
file.
- You're all set up! Mine your own data and don't hesitate to send
us your feedback and comments!
File format instructions
Please read carefully the following paragraph to avoid unnecessary exchanges.
- Data format: your data should be stored in a table, with records
in rows and fields in columns. The first row should hold the fields'
name.
- Data type: your data may be of any of the following types:
- Numerical (continuous or categorized, integer or real). Ex.:
Age (from 7 to 96)
- Ordered Symbolic. Ex.: Size (small, average, tall)
- Unordered symbolic. Ex.: Color (blue, green, red, yellow)
- Date, Time & DateTime. Ex.: Date of sale (MM/DD/YY)
- File format: your data file may be of any of the following formats:
Excel, Lotus, Access, Paradox, Foxpro, DBase, SAS, SPSS, Text (delimited
or fixed length).
- Data Size: There is no limit to the number of records, fields or
field values to be mined with ALICE d'ISoft.
However, as far as the "Mine your own data" offer is concerned, we can
only accept files that do not exceed a size of 70 Kb once zipped
(Be sure to turn the Max-compression option on!).
Example
Here is an example of what your file should look like:
| Customer Id |
Age |
Sex |
Job |
State |
Amount purchased |
Day of Sale |
Vendor |
Satisfied |
| Smith |
28 |
Male |
Firemen |
CA |
17 000 |
12/5/95 |
1 |
Yes |
| John |
47 |
Female |
Nurse |
NY |
9 500 |
2/28/96 |
1 |
Yes |
| O'Brian |
43 |
Male |
Professor |
NY |
21 000 |
6/15/96 |
2 |
No |
| Wayne |
25 |
Male |
Professor |
ME |
10 000 |
6/1/97 |
2 |
Yes |
| Lopez |
32 |
Female |
CEO |
CA |
35 000 |
3/6/95 |
2 |
No |
| Etc, etc. 100 000 more rows! |
Here, the field to be explained could either be "Amount Purchased" or
"Satisfied":
With "Satisfied", discover the factors which improve customer satisfaction,
With "Amount purchased", Find the most profitable customers' profile.
The field "Vendor" is coded using 1 and 2. These figures should only be
considered as identifiers and not as real numbers. The file description
for this file should therefore mention: "Vendor: Unordered Symbolic".
top
|
