Text Mining, AI, And WEKA Analysis Are Crucial For Essay Composition.

Question 1

What is text mining and how does it differ from data mining? Describe the technologies that can be used to process text mining and provide two applications of text mining .

Question 2

Understand how artificial intelligence (AI) is impacting business in some specific fields, such as healthcare, social media, supermarket and so on. Discuss the following topics:

• Why artificial intelligence is important for supporting business to build smart systems
• How artificial intelligence helps to transform companies
• The thread of artificial intelligence and the limitations of smart systems

Question 3

Read the resource material available on the course website about WEKA (in Week 5 and Week 6). Download WEKA software and install it onto your computer (ensure the bundled Java Runtime Environment, i.e. jre is installed). After successful installation of the program, classify bank accounts from the data bank-data.arff using J48 classifier (keep other parameters as default,). The bank-data.arff file can be found on the course website. Write an analysis report based on the classification results. You need to include the results in your report.

• To classify data in WEKA, use “Open file”, than select “Classify” ? “Classifier (Choose)” ? “trees” ?“J48”
• You need to develop a decision tree and explain the output as well as the important variables required for generating the output
• The report should include

oyour understanding of WEKA software and data analysis

oanalysis of classification results

osummary or conclusion

What is text mining and how it differs from data mining?

Text Analytics, otherwise called text mining, is the way toward looking at huge accumulations of composed assets to create new data, and to change the unstructured text into organized information for use in encourage investigation. Text mining distinguishes realities, connections and declarations that would some way or another stay covered in the mass of printed huge information. These certainties are removed and transformed into organized information, for investigation, representation (e.g. through html tables, mind maps, graphs), joining with organized information in databases or distribution centres, and further refinement utilizing machine learning (ML) frameworks.

Data mining is centredon data subordinate exercises, for example, bookkeeping, buying, production network, CRM, and so on. The required data is anything but difficult to get to and homogeneous. When calculations are characterized, the arrangement can be immediately conveyed. The multifaceted nature of the data prepared make text mining ventures longer to convey. Text mining checks a few go-between semantic phases of examination before it can improve text (dialect speculating, tokenization, division, morpho-syntactic investigation, disambiguation, cross-references, and so forth) (Mitsa, 2010). Next, significant terms extraction and metadata affiliation steps handle organizing the unstructured substance to support area particular applications. Additionally, tasks may include some heterogeneous dialects, organizations or spaces. At long last, few organizations have their own scientific categorization. Nonetheless, this is compulsory for beginning a text mining venture and it can take a couple of months to be created

Technologies and tools required for text mining

1. Sentiment analysis tool
2. Topic modelling technique
3. Named entity recognition and event extraction technique

Applications

1 – Risk administration

Regardless of the business, insufficient hazard investigation is regularly a main source of disappointment. This is particularly valid in the budgetary business where reception of Risk Management Software in view of text mining innovation can drastically expand the capacity to moderate hazard, empowering complete administration of thousands of sources and petabytes of text reports, and giving the capacity to interface together data and have the capacity to get to the correct data at the opportune time.

2 – Knowledge administration

Not having the capacity to discover imperative data rapidly is dependably a test while overseeing vast volumes of text reports—simply ask anybody in the human services industry. Here, associations are tested with a gigantic measure of data—many years of research in genomics and sub-atomic methods, for instance, and in addition volumes of clinical patient information—that could possibly be helpful for their biggest benefit focus: new item improvement. Here, learning administration programming in light of text mining offer an unmistakable and solid answer for the "data excess" issue.

3 – Cybercrime counteractive action

The unknown idea of the web and the numerous correspondence highlights worked through it add to the expanded danger of web based wrongdoings. Today, text mining knowledge and against wrongdoing applications are making web wrongdoing counteractive action less demanding for any venture and law implementation or insight organizations.

4 – Customer mind benefit

Text mining, and characteristic dialect handling are visit applications for client mind. Today, text investigation programming is much of the time received to enhance client encounter utilizing distinctive wellsprings of important data, for example, reviews, inconvenience tickets, and client call notes to enhance the quality, viability and speed in settling issues. Text examination is utilized to give a quick, computerized reaction to the client, drastically lessening their dependence accessible if the need arises focus administrators to take care of issues (Witten, Frank, Hall & Pal, 2017).

Technologies and tools required for text mining

Numerous organizations take up Artificial Intelligence (AI) innovation to attempt to diminish operational costs, increment productivity, develop income and enhance client encounter.

For most prominent advantages, organizations should take a gander at putting the full scope of brilliant innovations - including machine learning, regular dialect preparing and then some - into their procedures and items. In any case, even organizations that are new to AI can receive significant benefits (Berlatsky, 2011).

AI affects business by conveying the correct AI innovation, your business may pick up capacity to:

• spare time and cash via computerizing routine procedures and assignments
• increment profitability and operational efficiencies
• settle on quicker business choices in light of yields from psychological advancements
• stay away from mix-ups and 'human blunder', gave that keen frameworks are set up appropriately
• utilize understanding to foresee client inclinations and offer them better, customized involvement
• mine huge measure of information to create quality leads and develop your client base
• accomplish cost funds, by upgrading your business, your workforce or your items
• increment income by recognizing and expanding deals openings
• develop aptitude by empowering examination and offering astute exhortation and support
• As per an ongoing Infosys think about, the fundamental main thrust for utilizing AI in business was contender advantage. From that point onward, the impetus originated from:
• an official drove choice
• a specific business, operational or specialized issue
• an interior investigation
• client request
• a surprising answer for issue
• a branch of another undertaking

• proposals and substance curation
• personalization of news sustains
• example and picture acknowledgment
• dialect acknowledgment - to process unstructured information from clients and deals prospects
• promotion focusing on and streamlined, ongoing offering
• information investigation and client division
• social semantics and feeling investigation
• robotized website composition
• prescient client benefit

These are just a portion of the cases of AI utilizes as a part of business. With the pace of advancement expanding, there will probably be considerably more to come sooner rather than later.

With the quick improvement of AI, various moral issues have sprung up. These include:

• the capability of computerization innovation to offer ascent to work misfortunes
• the need to redeploy or retrain representatives to keep them in occupations
• reasonable dissemination of riches made by machines
• the impact of machine association on human conduct and consideration
• the need to dispose of inclination in AI that is made by people
• the security of AI frameworks (eg self-sufficient weapons) that can conceivably cause harm
• the need to relieve against unintended outcomes, as savvy machines are thought to learn and grow autonomously

While these dangers can't be disregarded, it merits remembering that advances in AI can - generally - make better business and better lives for everybody. On the off chance that actualized capably, computerized reasoning has gigantic and helpful potential (Ennals, 2014).

Here, we will do the data mining analysis for bank data. The bank data contains the bank data. The bank data analysis uses the J48 analysis in Weka data mining tool. Then, Do data mining analysis by using the below steps and it shown below (Sche?mas commente?s en sante? se?curite? au travail, 2011).

First, Open Weka data mining tool. It is shown below

It is shown below.

Once successfully load the data.

After, clicks classify tab and click choose to select the trees.

Then, click the J48 to do the J48 analysis.

The J48 analysis for each attributes is shown below.

J48 Analysis for ID

J48 Analysis for Income

Correctly Classified Instances         228               38      %

Incorrectly Classified Instances       372               62      %

Kappa statistic                          0.0233

Mean absolute error                      0.3326

Root mean squared error                  0.4748

Relative absolute error                 97.8135 %

Root relative squared error            115.1822 %

Total Number of Instances              600     

=== Detailed Accuracy By Class ===

                 TP Rate  FP Rate  Precision  Recall   F-Measure  MCC      ROC Area  PRC Area  Class

                 0.617    0.610    0.451      0.617    0.521      0.007    0.495     0.447     INNER_CITY

                 0.243    0.244    0.288      0.243    0.263      -0.001   0.517     0.291     TOWN

                 0.188    0.075    0.321      0.188    0.237      0.141    0.576     0.246     RURAL

                 0.032    0.052    0.067      0.032    0.043      -0.028   0.522     0.108     SUBURBAN

Weighted Avg.    0.380    0.361    0.343      0.380    0.352      0.023    0.517     0.335     

=== Confusion Matrix ===

a   b   c   d   <-- classified as

166  70  19  14 |   a = INNER_CITY

110  42  13   8 |   b = TOWN

53  19  18   6 |   c = RURAL

39  15   6   2 |   d = SUBURBA

J48 Analysis for Save_Act

Correctly Classified Instances         431               71.8333 %

Incorrectly Classified Instances       169               28.1667 %

Kappa statistic                          0.2968

Mean absolute error                      0.3197

Root mean squared error                  0.4565

Relative absolute error                 74.6875 %

Root relative squared error             98.7075 %

Total Number of Instances              600     

=== Detailed Accuracy By Class ===

                 TP Rate  FP Rate  Precision  Recall   F-Measure  MCC      ROC Area  PRC Area  Class

Applications of text mining

                 0.430    0.152    0.559      0.430    0.486      0.302    0.738     0.504     NO

                 0.848    0.570    0.768      0.848    0.806      0.302    0.738     0.858     YES

Weighted Avg.    0.718    0.440    0.703      0.718    0.707      0.302    0.738     0.748     

=== Confusion Matrix ===

a   b   <-- classified as

  80 106 |   a = NO

  63 351 |   b = YES

J48 Analysis for Mortgage

Correctly Classified Instances         393               65.5    %

Incorrectly Classified Instances       207               34.5    %

Kappa statistic                          0.1149

Mean absolute error                      0.4169

Root mean squared error                  0.4904

Relative absolute error                 91.8046 %

Root relative squared error            102.9264 %

Total Number of Instances              600     

=== Detailed Accuracy By Class ===

                 TP Rate  FP Rate  Precision  Recall   F-Measure  MCC      ROC Area  PRC Area  Class

                 0.898    0.799    0.678      0.898    0.772      0.137    0.568     0.704     NO

                 0.201    0.102    0.512      0.201    0.289      0.137    0.568     0.424     YES

Weighted Avg.    0.655    0.556    0.620      0.655    0.604      0.137    0.568     0.607     

=== Confusion Matrix ===

a   b   <-- classified as

351  40 |   a = NO

167  42 |   b = YES

J48 Analysis for Sex

=== Stratified cross-validation ===

Correctly Classified Instances         321               53.5    %

Incorrectly Classified Instances       279               46.5    %

Kappa statistic                          0.07  

Mean absolute error                      0.4994

Root mean squared error                  0.5876

Relative absolute error                 99.8804 %

Root relative squared error            117.5249 %

Total Number of Instances              600     

=== Detailed Accuracy by Class === TP Rate  FP Rate  Precision  Recall   F-Measure  MCC      ROC Area  PRC Area  Class

0.493    0.423    0.538      0.493    0.515      0.070    0.494     0.491     FEMALE

0.577    0.507    0.532      0.577    0.554      0.070    0.494     0.487     MALE

Weighted Avg.    0.535    0.465    0.535      0.535    0.534      0.070    0.494     0.489     

=== Confusion Matrix ===

a   b   <-- classified as

 148 152 |   a = FEMALE

 127 173 |   b = MALE

J48 Analysis for Save_Act

Correctly Classified Instances         455               75.8333 %

Incorrectly Classified Instances       145               24.1667 %

Kappa statistic                          0     

Mean absolute error                      0.3665

Root mean squared error                  0.4281

Relative absolute error                 99.8658 %

Root relative squared error             99.9998 %

Total Number of Instances              600     

=== Detailed Accuracy By Class ===

                 TP Rate  FP Rate  Precision  Recall   F-Measure  MCC      ROC Area  PRC Area  Class

                 0.000    0.000    0.000      0.000    0.000      0.000    0.489     0.238     NO

                 1.000    1.000    0.758      1.000    0.863      0.000    0.489     0.754     YES

Weighted Avg.    0.758    0.758    0.575      0.758    0.654      0.000    0.489     0.629     

=== Confusion Matrix ===

a   b   <-- classified as

   0 145 |   a = NO

   0 455 |   b = YES

The above dashboard is outlined below.

A Sales Cycle is an impression of the time it takes to go through and through, from potential chance to make a deal to a handshake and a paid receipt. In any case, amidst the invigoration of making the deal, subtle elements get lost, efficiency isn't generally at its most astounding and we won't not work as beneficially as we picture or expectation. Clearly, shortening the Sales Cycle can fundamentally affect our main concern, put more cash in the bank and guarantee a more joyful and more dedicated client base. Notwithstanding, before we can upgrade the cycle, we should first comprehend it. Enter our business dashboard programming and assemble our individual deals cycle length dashboard. We can begin with the fundamentals, by first taking a gander at our present deals cycle length after some time to use as a benchmark (Check-listes pour cadres dirigeants performants, 2012).

In this business dashboard layout, the length of a Sales Cycle is delineated as a business pipe with four stages in making a deal; Opportunities, Proposals, Negotiations and Closings. Each progression in this procedure takes a specific measure of time and the normal length of the Sales Cycle is an impression of the normal time each stage should be finished, over all delegates inside our business group.

In the wake of setting up and tweaking our own particular Sales Cycle stages, in accordance with our business and specific item stock, we can begin evaluating our business efficiency and that of every individual deals agent also. Is it true that one is rep fundamentally beating other colleagues? Since we have that data readily available, we can bore down to discover what is and isn't working. We can likewise utilize this business dashboard to track singular rep's advance after some time. This assists with objective setting for people, and for the business group all in all.

We definitely realize that the primary concern is to expand benefits and enhance profitability; utilizing cutting edge Sales Cycle Length perceptions will help we promptly decide the qualities and shortcomings in our business group and activities, and give we the data we have to react as needs be.

References

Berlatsky, N. (2011). Artificial intelligence. Detroit: Greenhaven Press.

Ennals, J. (2014). Artificial Intelligence. Elsevier Science.

Mitsa, T. (2010). Temporal data mining. Boca Raton, FL: Chapman & Hall/CRC.

Tissot. (2011). Sche?mas commente?s en sante? se?curite? au travail. Annecy-le-Vieux.

WEKA Business Media. (2012). Check-listes pour cadres dirigeants performants. Zurich.

Witten, I., Frank, E., Hall, M., & Pal, C. (2017). Data mining. Amsterdam: Morgan Kaufmann.