SAP HANA Predictive Analysis Library (PAL)

Recommendations

Info

DROP TABLE PAL_AP_SEED_TBL; CREATE COLUMN TABLE PAL_AP_SEED_TBL LIKE PAL_AP_SEED_T; DROP TABLE #PAL_CONTROL_TBL; CREATE LOCAL TEMPORARY COLUMN TABLE #PAL_CONTROL_TBL LIKE PAL_CONTROL_T; INSERT INTO #PAL_CONTROL_TBL VALUES('THREAD_NUMBER',2,null,null); INSERT INTO #PAL_CONTROL_TBL VALUES('MAX_ITERATION',500,null,null); INSERT INTO #PAL_CONTROL_TBL VALUES('CON_ITERATION',100,null,null); INSERT INTO #PAL_CONTROL_TBL VALUES('DAMP',null,0.9,null); INSERT INTO #PAL_CONTROL_TBL VALUES('PREFERENCE',null,0.5,null); INSERT INTO #PAL_CONTROL_TBL VALUES('DISTANCE_METHOD',2,null,null); INSERT INTO #PAL_CONTROL_TBL VALUES('CLUSTER_NUMBER',0,null,null); DROP TABLE PAL_AP_RESULTS_TBL; CREATE COLUMN TABLE PAL_AP_RESULTS_TBL LIKE PAL_AP_RESULTS_T; CALL DM_PAL.PAL_AP(PAL_AP_DATA_TBL, PAL_AP_SEED_TBL, #PAL_CONTROL_TBL, PAL_AP_RESULTS_TBL) with OVERVIEW; SELECT * FROM PAL_AP_RESULTS_TBL; Expected Result 24 P U B L I C SAP HANA Predictive Analysis Library (PAL) PAL Functions
3.1.2 Agglomerate Hierarchical Clustering Hierarchical clustering is a widely used clustering method which can find natural groups within a set of data. The idea is to group the data into a hierarchy or a binary tree of the subgroups. A hierarchical clustering can be either agglomerate or divisive, depending on the method of hierarchical decomposition. The implementation in PAL follows the agglomerate approach, which merges the clusters with a bottom-up strategy. Initially, each data point is considered as an own cluster. The algorithm iteratively merges two clusters based on the dissimilarity measure in a greedy manner and forms a larger cluster. Therefore, the input data must be numeric and a measure of dissimilarity between sets of data is required, which is achieved by using the following two parameters: ● ● An appropriate metric (a measure of distance between pairs of groups) A linkage criterion which specifies the distances between groups An advantage of hierarchical clustering is that it does not require the number of clusters to be specified as the input. And the hierarchical structure can also be used for data summarization and visualization. The agglomerate hierarchical clustering functions in PAL now supports eight kinds of appropriate metrics and seven kinds of linkage criteria. Support for Category Attributes If the input data has category attributes, you must set the DISTANCE_FUNC parameter to Gower Distance to support calculating the distance matrix. Gower Distance is calculated in the following way. Suppose that the items X i and X j have K attributes, the distance between X i and X j is: For continuous attributes, S ijk =(X ik ‒X jk )/R k W k =1 R k is the range of values for the k th variable; W k is set by user and the default is 1. For category attributes, If X ik =X jk : S ijk =0 Other cases: S ijk =1 Prerequisites ● ● The first column of the input data is an ID column and the other columns are of integer, double, varchar, or nvarchar data type. The input data does not contain null value. The algorithm will issue errors when encountering null values. SAP HANA Predictive Analysis Library (PAL) PAL Functions P U B L I C 25
Page 1 and 2: SAP HANA Platform SPS 12 Document V
Page 3 and 4: Naive Bayes. ......................
Page 5 and 6: 1 What is PAL? SAP HANA’s SQLScri
Page 7 and 8: 2 Getting Started with PAL This sec
Page 9 and 10: The result will tell you whether th
Page 11 and 12: ● : The name of the schema where
Page 13 and 14: CREATE COLUMN TABLE PAL_ARIMAX_DATA
Page 15 and 16: For more information on how to use
Page 17 and 18: Category PAL Algorithm Built-in Fun
Page 19 and 20: Category PAL Algorithm Built-in Fun
Page 21 and 22: The procedure name is the same as s
Page 23: ● USER1 has been assigned the AFL
Page 27 and 28: Table 13: Name Data Type Default Va
Page 29 and 30: Name Data Type Default Value Descri
Page 31 and 32: INSERT INTO DATA_TBL VALUES ('17',
Page 33 and 34: Prerequisites ● ● The input dat
Page 37 and 38: INSERT INTO PAL_AD_DATA_TBL VALUES
Page 39 and 40: The signature table should contain
Page 41 and 42: CREATE TYPE PAL_KMEANS_CENTERS_T AS
Page 43 and 44: PAL_CLUSTER_ASSIGNED_TBL: For DBSCA
Page 45 and 46: For SOM: SET SCHEMA DM_PAL; DROP TY
Page 47 and 48: PAL_CLUSTER_ASSIGNED_TBL: Related I
Page 49 and 50: Table 24: Name Data Type Descriptio
Page 51 and 52: DROP TYPE PAL_CONTROL_T; CREATE TYP
Page 53 and 54: 3.1.6 Gaussian Mixture Model (GMM)
Page 55 and 56: Table 29: Name Data Type Descriptio
Page 57 and 58: Output Tables Output format 0: Tabl
Page 59 and 60: INSERT INTO PAL_GMM_PDATA_TBL VALUE
Page 61 and 62: Example 2 SET SCHEMA DM_PAL; DROP T
Page 63 and 64: PAL_GMM_RESULTSMODEL_TBL: 3.1.7 K-M
Page 65 and 66: The signature table should contain
Page 69 and 70: Table 43: Table Column Column Data
Page 71 and 72: INSERT INTO #PAL_CONTROL_TBL VALUES
Page 73 and 74: PAL_KMEANS_MODEL_TBL: VALIDATEKMEAN
Page 75 and 76:
"V000" DOUBLE, "A0" INTEGER, "A1" I
Page 77 and 78:
Expected Result PAL_SILHOUETTE_RESU
Page 79 and 80:
Table Column Column Data Type Descr
Page 81 and 82:
Example Assume that: ● ● ● DM
Page 83 and 84:
PAL_KMEDIANS_ASSIGN_TBL: 3.1.9 K-Me
Page 85 and 86:
Position Schema Name Table Type Nam
Page 87 and 88:
Name Data Type Default Value Descri
Page 89 and 90:
INSERT INTO #PAL_CONTROL_TBL VALUES
Page 91 and 92:
The input, parameter, and output ta
Page 93 and 94:
Table 67: Table Column Column Data
Page 95 and 96:
INSERT INTO PARAMETERS_TBL VALUES (
Page 97 and 98:
DOCTOPICDIST_TBL: GENERALINFO_TBL:
Page 99 and 100:
Procedure Calling CALL .(, , ) with
Page 101 and 102:
Page 103 and 104:
Expected Result DOCTOPICDIST_TBL: 3
Page 105 and 106:
The following parameters are option
Page 107 and 108:
Page 109 and 110:
Page 111 and 112:
CREATE TYPE PAL_SOM_MODEL_T AS TABL
Page 113 and 114:
PAL_SOM_RESASSIGN_TBL: PAL_SOM_MODE
Page 115 and 116:
Page 117 and 118:
INSERT INTO PAL_SLIGHT_SIL_PDATA_TB
Page 119 and 120:
AUC Procedure Generation CALL SYS.A
Page 121 and 122:
Table 85: Name Data Type Default Va
Page 123 and 124:
Example 2 SET SCHEMA DM_PAL; DROP T
Page 125 and 126:
3.2.2 Back Propagation Neural Netwo
Page 127 and 128:
CREATEBPNN This function trains a B
Page 129 and 130:
Page 131 and 132:
INSERT INTO PAL_TRAIN_NN_DATA_TBL V
Page 133 and 134:
PAL_TRAIN_NN_RESULT_TBL: PAL_REGRES
Page 135 and 136:
Classification example: SET SCHEMA
Page 137 and 138:
"NAME" VARCHAR(100), "MODEL" CLOB )
Page 139 and 140:
The signature table should contain
Page 141 and 142:
Page 143 and 144:
CREATE TYPE PAL_C45_DATA_T AS TABLE
Page 145 and 146:
CART This function is used for clas
Page 147 and 148:
Page 149 and 150:
Page 151 and 152:
Prerequisites ● ● The target co
Page 153 and 154:
Page 155 and 156:
Page 157 and 158:
PAL_CHAID_TREEMODEL_TBL: PAL_CHAID_
Page 159 and 160:
Page 161 and 162:
INSERT INTO PAL_CM_DATA_TBL VALUES(
Page 163 and 164:
Page 165 and 166:
PAL_KNN_RESULTS_TBL: 3.2.8 Logistic
Page 167 and 168:
Procedure Calling CALL .(, , , , )
Page 169 and 170:
Page 171 and 172:
Page 173 and 174:
PAL_LOGISTICR_STAT_TBL: PAL_LOGISTI
Page 175 and 176:
Procedure Generation CALL SYS.AFLLA
Page 177 and 178:
Assume that: ● ● ● DM_PAL is
Page 179 and 180:
PAL_FLOGISTICR_FITTED_TBL: Related
Page 181 and 182:
Page 183 and 184:
Page 185 and 186:
Expected Result PAL_LRMC_MODEL_TBL:
Page 187 and 188:
Table 137: Name Data Type Default V
Page 189 and 190:
SAP HANA Predictive Analysis Librar
Page 191 and 192:
CALL SYS.AFLLANG_WRAPPER_PROCEDURE_
Page 193 and 194:
3.2.10 Naive Bayes Naive Bayes is a
Page 195 and 196:
Page 197 and 198:
● ● USER1 has been assigned the
Page 199 and 200:
Page 201 and 202:
"DOUBLEARGS" DOUBLE, "STRINGARGS" V
Page 203 and 204:
Position Schema Name Table Type Nam
Page 205 and 206:
Page 207 and 208:
Page 209 and 210:
DROP TABLE CV_LR_FUNC_TBL; CREATE C
Page 211 and 212:
CREATE COLUMN TABLE FUNC_TBL LIKE F
Page 213 and 214:
DROP TABLE EVALUATION_RESULT_TBL; C
Page 215 and 216:
INSERT INTO PAL_PARA_TBL VALUES('TR
Page 217 and 218:
Page 219 and 220:
3.2.13 Random Forest The ransom for
Page 221 and 222:
Page 223 and 224:
INSERT INTO PAL_RF_DATA_TBL VALUES
Page 225 and 226:
Page 227 and 228:
3.2.14 Support Vector Machine Suppo
Page 229 and 230:
Table 168: Name Data Type Descripti
Page 231 and 232:
Page 233 and 234:
CREATE TABLE PAL_SVM_PDATA_TBL("POS
Page 235 and 236:
Page 237 and 238:
STRINGARGS VARCHAR(100) ); DROP TYP
Page 239 and 240:
Example 3: Support vector classific
Page 241 and 242:
Related Information SAP HANA Smart
Page 243 and 244:
Page 245 and 246:
INSERT INTO PAL_GR_PDATA_TBL values
Page 247 and 248:
Procedure Calling CALL .(, , , ) wi
Page 249 and 250:
3.3.2 Bi-Variate Natural Logarithmi
Page 251 and 252:
Page 253 and 254:
CREATE LOCAL TEMPORARY COLUMN TABLE
Page 255 and 256:
Page 257 and 258:
Where β0…βn are parameters that
Page 259 and 260:
Page 261 and 262:
Page 263 and 264:
Page 265 and 266:
Assume we have m observation pairs
Page 267 and 268:
Page 269 and 270:
Page 271 and 272:
Example 1: Fitting multiple linear
Page 273 and 274:
CREATE COLUMN TABLE PAL_ENET_MLR_PD
Page 275 and 276:
Page 277 and 278:
3.3.5 Polynomial Regression Polynom
Page 279 and 280:
Page 281 and 282:
insert into PAL_PR_PDATA_TBL values
Page 283 and 284:
The input, parameter, and output ta
Page 285 and 286:
3.4 Association Algorithms This sec
Page 287 and 288:
Page 289 and 290:
Page 291 and 292:
"STRINGARGS" VARCHAR (100) ); INSER
Page 293 and 294:
Optional Parameters The following p
Page 295 and 296:
Page 297 and 298:
INSERT INTO PAL_APRIORI_DATA_TBL VA
Page 299 and 300:
PAL_APRIORI_CONS_ITEMS_TBL: LITEAPR
Page 301 and 302:
Page 303 and 304:
INSERT INTO PAL_LITEAPRIORI_DATA_TB
Page 305 and 306:
Page 307 and 308:
Page 309 and 310:
FPGROWTH (Relational Output) FP-Gro
Page 311 and 312:
Page 313 and 314:
); DROP TYPE PAL_FPGROWTH_POSTRULE_
Page 315 and 316:
PAL_FPGROWTH_VALUES_TBL: 3.4.3 K-Op
Page 317 and 318:
Page 319 and 320:
INSERT INTO PAL_KORD_DATA_TBL VALUE
Page 321 and 322:
Prerequisite No missing or null dat
Page 323 and 324:
. Initial P 0 can be calculated by
Page 325 and 326:
Output Table (Model Table) Table 24
Page 327 and 328:
DROP TYPE PAL_ARIMAX_DATA_T; CREATE
Page 329 and 330:
CALL "SYS".AFLLANG_WRAPPER_PROCEDUR
Page 331 and 332:
We get And variance of the forecast
Page 333 and 334:
CREATE COLUMN TABLE PAL_ARIMA_PDATA
Page 335 and 336:
Page 337 and 338:
INSERT INTO PAL_ARIMAX_DATA_TBL VAL
Page 339 and 340:
3.5.2 Auto ARIMA This function auto
Page 341 and 342:
Page 343 and 344:
Page 345 and 346:
● USER1 has been assigned the AFL
Page 347 and 348:
Page 349 and 350:
Example 3: Auto Seasonal ARIMA SET
Page 351 and 352:
INSERT INTO PAL_ARIMA_DATA_TBL VALU
Page 353 and 354:
INSERT INTO PAL_ARIMA_DATA_TBL VALU
Page 355 and 356:
F 1 = a 0 + b 0 2. Calculation: S t
Page 357 and 358:
Page 359 and 360:
INSERT INTO PAL_BROWNSMOOTH_DATA_TB
Page 361 and 362:
3.5.4 Croston's Method The Croston
Page 363 and 364:
Page 365 and 366:
SELECT * FROM PAL_CROSTON_STATISTIC
Page 367 and 368:
Page 369 and 370:
PAL_FORECASTACCURACYMEASURES_RESULT
Page 371 and 372:
Page 373 and 374:
Page 375 and 376:
Page 377 and 378:
PAL_OUTPARAMETER_TBL: PAL_FORECASTS
Page 379 and 380:
PAL_OUTPARAMETER_TBL: SAP HANA Pred
Page 381 and 382:
CREATE TYPE PAL_CONTROL_T AS TABLE
Page 383 and 384:
PAL_FORECASTTRIPLESMOOTHING_RESULT_
Page 385 and 386:
PAL_OUTPARAMETER_TBL: PAL_FORECASTM
Page 387 and 388:
PAL_OUTPARAMETER_TBL: PAL_FORECASTM
Page 389 and 390:
Procedure Calling CALL .(, , , ) wi
Page 391 and 392:
Page 393 and 394:
PAL_FORECASTSLR_RESULT_TBL: PAL_FOR
Page 395 and 396:
Note The algorithm is backward comp
Page 397 and 398:
Page 399 and 400:
Page 401 and 402:
Note F 0 is not defined because the
Page 403 and 404:
Page 405 and 406:
CREATE COLUMN TABLE PAL_DOUBLESMOOT
Page 407 and 408:
3.5.10 Triple Exponential Smoothing
Page 409 and 410:
PAL calculates the prediction inter
Page 411 and 412:
Page 413 and 414:
Output Table Table 294: Table Colum
Page 415 and 416:
PAL_TRIPLESMOOTH_RESULT_TBL: PAL_TR
Page 417 and 418:
Once the trend and seasonality are
Page 419 and 420:
Page 421 and 422:
CREATE COLUMN TABLE PAL_TSSEASONALI
Page 423 and 424:
The difference-sign test counts the
Page 425 and 426:
Optional Parameters The following p
Page 427 and 428:
PAL_TSTREND_TREND_TBL: PAL_TSTREND_
Page 429 and 430:
Parameter Table Mandatory Parameter
Page 431 and 432:
3.6 Preprocessing Algorithms The re
Page 433 and 434:
● ● Smoothing by bin medians: e
Page 435 and 436:
Page 437 and 438:
3.6.2 Binning Assignment Binning as
Page 439 and 440:
Page 441 and 442:
3.6.3 Convert Category Type to Bina
Page 443 and 444:
"NAME" VARCHAR(100), "INTARGS" INTE
Page 445 and 446:
Procedure Generation CALL SYS.AFLLA
Page 447 and 448:
INSERT INTO PAL_IQR_TESTDT_TBL VALU
Page 449 and 450:
PARTITION This function reads the i
Page 451 and 452:
Page 453 and 454:
3.6.6 Posterior Scaling Posterior s
Page 455 and 456:
Page 457 and 458:
Expected Result PAL_NEW_SCALING_TBL
Page 459 and 460:
Page 461 and 462:
INSERT INTO PAL_PCA_DATA_TBL VALUES
Page 463 and 464:
Page 465 and 466:
INSERT INTO PAL_PCAPROJ_PDATA_TBL V
Page 467 and 468:
Page 469 and 470:
INSERT INTO PAL_DISTRRANDOM_PDATA_T
Page 471 and 472:
Page 473 and 474:
INSERT INTO PAL_SAMPLING_DATA_TBL V
Page 475 and 476:
If method is 4 and SAMPLING_SIZE is
Page 477 and 478:
DROP TABLE PAL_SAMPLING_PDATA_TBL;
Page 479 and 480:
Expected Result 3.6.10 Scaling Rang
Page 481 and 482:
Page 483 and 484:
Page 485 and 486:
Prerequisite Each column must conta
Page 487 and 488:
Page 489 and 490:
PAL_MISSING_VALUES_RESULT_TBL: PAL_
Page 491 and 492:
The procedure name is the same as s
Page 493 and 494:
INSERT INTO PAL_VT_DATA_TBL VALUES
Page 495 and 496:
The procedure name is the same as s
Page 497 and 498:
3.7.2 Chi-Squared Test for Independ
Page 499 and 500:
Page 501 and 502:
DISTRPROB This function calculates
Page 503 and 504:
Page 505 and 506:
Page 507 and 508:
"VALUE" DOUBLE ); DROP TABLE PDATA_
Page 509 and 510:
Page 511 and 512:
"DOUBLEARGS" DOUBLE, "STRINGARGS" V
Page 513 and 514:
"TYPE_NAME" NVARCHAR(256), "PARAMET
Page 515 and 516:
2. Calculate PAL also supports the
Page 517 and 518:
Page 519 and 520:
The Kaplan-Meier estimator can be r
Page 521 and 522:
Page 523 and 524:
Page 525 and 526:
PAL_KMSURV_LOGRANK_STAT2_TBL: 3.7.7
Page 527 and 528:
Page 529 and 530:
Page 531 and 532:
CREATE TYPE PAL_DISTRQUANTILE_RESUL
Page 533 and 534:
where x' = x - Table 404: Definitio
Page 535 and 536:
Page 537 and 538:
Let the sample mean of x and y be:
Page 539 and 540:
CREATE TYPE PAL_VAREQUALTEST_DATA1_
Page 541 and 542:
The quantity is computed as a weigh
Page 543 and 544:
Page 545 and 546:
Prerequisites ● ● Input data ca
Page 547 and 548:
"PARAMETER_TYPE" VARCHAR(7) ); INSE
Page 549 and 550:
Signature Input Tables Table 425: T
Page 551 and 552:
CALL "SYS".AFLLANG_WRAPPER_PROCEDUR
Page 553 and 554:
"DISTANCE" DOUBLE ); DROP TYPE PAL_
Page 555 and 556:
"STRINGARGS" VARCHAR(100) ); --crea
Page 557 and 558:
The projected estimates are single
Page 559 and 560:
INSERT INTO PDATA VALUES (2, 'DM_PA
Page 561 and 562:
----product revenue---- DELETE FROM
Page 563 and 564:
---- calculate net present value of
Page 565 and 566:
UPDATE PAL_DISTRPROB_DISTRPARAM_TBL
Page 567 and 568:
Then we get the survival estimates
Page 569 and 570:
INSERT INTO PAL_TRIAL_DATA_TBL VALU
Page 571 and 572:
INSERT INTO PAL_DISTRFITCENSORED_DA
Page 573 and 574:
The expected results are as follows
Page 575 and 576:
); INSERT INTO PAL_DISTRPROB_PDATA_
Page 577 and 578:
Important Disclaimers and Legal Inf
show all

SAP HANA Predictive Analysis Library (PAL)

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?