int main()
{
char str[] = "string";
printf ("%d", findRank(str));
return 0;
}
int main(int argc, char **argv)
{
int iArg;
char **column, **excludeColumn, *withOnly;
long columns, excludeColumns;
char *input, *output;
SCANNED_ARG *scanned;
SDDS_DATASET SDDSin, SDDSout;
long i, j, row, rows, count, readCode, rankOrder, iName1, iName2;
int32_t outlierStDevPasses;
double **data, correlation, significance, outlierStDevLimit;
double **rank;
short **accept;
char s[SDDS_MAXLINE];
unsigned long pipeFlags, dummyFlags, majorOrderFlag;
short columnMajorOrder=-1;
SDDS_RegisterProgramName(argv[0]);
argc = scanargs(&scanned, argc, argv);
if (argc<2)
bomb(NULL, USAGE);
output = input = withOnly = NULL;
columns = excludeColumns = 0;
column = excludeColumn = NULL;
pipeFlags = 0;
rankOrder = 0;
outlierStDevPasses = 0;
outlierStDevLimit = 1.;
rank = NULL;
accept = NULL;
for (iArg=1; iArg<argc; iArg++) {
if (scanned[iArg].arg_type==OPTION) {
/* process options here */
switch (match_string(scanned[iArg].list[0], option, N_OPTIONS, 0)) {
case SET_MAJOR_ORDER:
majorOrderFlag=0;
scanned[iArg].n_items--;
if (scanned[iArg].n_items>0 &&
(!scanItemList(&majorOrderFlag, scanned[iArg].list+1, &scanned[iArg].n_items, 0,
"row", -1, NULL, 0, SDDS_ROW_MAJOR_ORDER,
"column", -1, NULL, 0, SDDS_COLUMN_MAJOR_ORDER,
NULL)))
SDDS_Bomb("invalid -majorOrder syntax/values");
if (majorOrderFlag&SDDS_COLUMN_MAJOR_ORDER)
columnMajorOrder=1;
else if (majorOrderFlag&SDDS_ROW_MAJOR_ORDER)
columnMajorOrder=0;
break;
case SET_COLUMNS:
if (columns)
SDDS_Bomb("only one -columns option may be given");
if (scanned[iArg].n_items<2)
SDDS_Bomb("invalid -columns syntax");
column = tmalloc(sizeof(*column)*(columns=scanned[iArg].n_items-1));
for (i=0; i<columns; i++)
column[i] = scanned[iArg].list[i+1];
break;
case SET_EXCLUDE:
if (scanned[iArg].n_items<2)
SDDS_Bomb("invalid -excludeColumns syntax");
moveToStringArray(&excludeColumn, &excludeColumns, scanned[iArg].list+1, scanned[iArg].n_items-1);
break;
case SET_WITHONLY:
if (withOnly)
SDDS_Bomb("only one -withOnly option may be given");
if (scanned[iArg].n_items<2)
SDDS_Bomb("invalid -withOnly syntax");
withOnly = scanned[iArg].list[1];
break;
case SET_PIPE:
if (!processPipeOption(scanned[iArg].list+1, scanned[iArg].n_items-1, &pipeFlags))
SDDS_Bomb("invalid -pipe syntax");
break;
case SET_RANKORDER:
rankOrder = 1;
break;
case SET_STDEVOUTLIER:
scanned[iArg].n_items--;
outlierStDevPasses = 1;
outlierStDevLimit = 1;
if (!scanItemList(&dummyFlags, scanned[iArg].list+1, &scanned[iArg].n_items, 0,
"limit", SDDS_DOUBLE, &outlierStDevLimit, 1, 0,
"passes", SDDS_LONG, &outlierStDevPasses, 1, 0,
NULL) ||
outlierStDevPasses<=0 || outlierStDevLimit<=0)
SDDS_Bomb("invalid -stdevOutlier syntax/values");
break;
default:
fprintf(stderr, "error: unknown/ambiguous option: %s\n",
scanned[iArg].list[0]);
exit(1);
break;
}
}
else {
if (!input)
input = scanned[iArg].list[0];
else if (!output)
output = scanned[iArg].list[0];
else
SDDS_Bomb("too many filenames seen");
}
}
processFilenames("sddscorrelate", &input, &output, pipeFlags, 0, NULL);
if (!SDDS_InitializeInput(&SDDSin, input))
SDDS_PrintErrors(stderr, SDDS_VERBOSE_PrintErrors|SDDS_EXIT_PrintErrors);
if (!columns)
columns = appendToStringArray(&column, columns, "*");
if (withOnly)
columns = appendToStringArray(&column, columns, withOnly);
if ((columns=expandColumnPairNames(&SDDSin, &column, NULL, columns,
excludeColumn, excludeColumns, FIND_NUMERIC_TYPE, 0))<=0) {
SDDS_PrintErrors(stderr, SDDS_VERBOSE_PrintErrors|SDDS_EXIT_PrintErrors);
SDDS_Bomb("no columns selected for correlation analysis");
}
if (!SDDS_InitializeOutput(&SDDSout, SDDS_BINARY, 0, NULL, "sddscorrelate output", output) ||
SDDS_DefineColumn(&SDDSout, "Correlate1Name", NULL, NULL, "Name of correlated quantity 1", NULL, SDDS_STRING, 0)<0 ||
SDDS_DefineColumn(&SDDSout, "Correlate2Name", NULL, NULL, "Name of correlated quantity 2", NULL, SDDS_STRING, 0)<0 ||
SDDS_DefineColumn(&SDDSout, "CorrelatePair", NULL, NULL, "Names of correlated quantities", NULL, SDDS_STRING, 0)<0 ||
SDDS_DefineColumn(&SDDSout, "CorrelationCoefficient", "r", NULL, "Linear correlation coefficient",
NULL, SDDS_DOUBLE, 0)<0 ||
SDDS_DefineColumn(&SDDSout, "CorrelationSignificance", "P$br$n", NULL, "Linear correlation coefficient significance",
NULL, SDDS_DOUBLE, 0)<0 ||
SDDS_DefineColumn(&SDDSout, "CorrelationPoints", NULL, NULL, "Number of points used for correlation",
NULL, SDDS_LONG, 0)<0 ||
SDDS_DefineParameter(&SDDSout, "CorrelatedRows", NULL, NULL, "Number of data rows in correlation analysis",
NULL, SDDS_LONG, NULL)<0 ||
SDDS_DefineParameter(&SDDSout, "sddscorrelateInputFile", NULL, NULL, "Data file processed by sddscorrelate",
NULL, SDDS_STRING, input?input:"stdin")<0 ||
SDDS_DefineParameter(&SDDSout, "sddscorrelateMode", NULL, NULL, NULL, NULL, SDDS_STRING,
rankOrder?"Rank-Order (Spearman)":"Linear (Pearson)")<0 ||
SDDS_DefineParameter1(&SDDSout, "sddscorrelateStDevOutlierPasses", NULL, NULL,
"Number of passes of standard-deviation outlier elimination applied",
NULL, SDDS_LONG, &outlierStDevPasses)<0 ||
SDDS_DefineParameter1(&SDDSout, "sddscorrelateStDevOutlierLimit", NULL, NULL,
"Standard-deviation outlier limit applied",
NULL, SDDS_DOUBLE, &outlierStDevLimit)<0)
SDDS_PrintErrors(stderr, SDDS_VERBOSE_PrintErrors|SDDS_EXIT_PrintErrors);
if (columnMajorOrder!=-1)
SDDSout.layout.data_mode.column_major = columnMajorOrder;
else
SDDSout.layout.data_mode.column_major = SDDSin.layout.data_mode.column_major;
if (!SDDS_WriteLayout(&SDDSout))
SDDS_PrintErrors(stderr, SDDS_VERBOSE_PrintErrors|SDDS_EXIT_PrintErrors);
if (!(data = (double**)malloc(sizeof(*data)*columns)) ||
(rankOrder && !(rank = (double**)malloc(sizeof(*rank)*columns))) ||
!(accept = (short**)malloc(sizeof(*accept)*columns)))
SDDS_Bomb("allocation failure");
while ((readCode=SDDS_ReadPage(&SDDSin))>0) {
if ((rows = SDDS_CountRowsOfInterest(&SDDSin))<3)
continue;
if (!SDDS_StartPage(&SDDSout, columns*(columns-1)/2) ||
!SDDS_SetParameters(&SDDSout, SDDS_SET_BY_NAME|SDDS_PASS_BY_VALUE,
"CorrelatedRows", rows, NULL))
SDDS_PrintErrors(stderr, SDDS_VERBOSE_PrintErrors|SDDS_EXIT_PrintErrors);
for (i=0; i<columns; i++) {
if (!(data[i] = SDDS_GetColumnInDoubles(&SDDSin, column[i])))
SDDS_PrintErrors(stderr, SDDS_VERBOSE_PrintErrors|SDDS_EXIT_PrintErrors);
if (rankOrder)
rank[i] = findRank(data[i], rows);
if (outlierStDevPasses) {
if (!(accept[i] = (short*)malloc(sizeof(**accept)*rows)))
SDDS_Bomb("allocation failure");
markStDevOutliers(data[i], outlierStDevLimit, outlierStDevPasses,
accept[i], rows);
}
else
accept[i] = NULL;
}
for (i=row=0; i<columns; i++) {
for (j=i+1; j<columns; j++) {
iName1 = i;
iName2 = j;
if (withOnly) {
if (strcmp(withOnly, column[i])==0) {
iName1 = j;
iName2 = i;
}
else if (strcmp(withOnly, column[j])==0) {
iName1 = i;
iName2 = j;
}
else
continue;
}
correlation
= linearCorrelationCoefficient(rankOrder?rank[i]:data[i],
rankOrder?rank[j]:data[j],
accept[i], accept[j], rows, &count);
significance = linearCorrelationSignificance(correlation, count);
sprintf(s, "%s.%s", column[iName1], column[iName2]);
if (!SDDS_SetRowValues(&SDDSout, SDDS_SET_BY_INDEX|SDDS_PASS_BY_VALUE, row++,
0, column[iName1],
1, column[iName2],
2, s, 3, correlation, 4, significance,
5, count, -1))
SDDS_PrintErrors(stderr, SDDS_VERBOSE_PrintErrors|SDDS_EXIT_PrintErrors);
}
}
for (i=0; i<columns; i++) {
free(data[i]);
if (rankOrder)
free(rank[i]);
if (accept[i])
free(accept[i]);
}
if (!SDDS_WritePage(&SDDSout))
SDDS_PrintErrors(stderr, SDDS_VERBOSE_PrintErrors|SDDS_EXIT_PrintErrors);
}
if (!SDDS_Terminate(&SDDSin)) {
SDDS_PrintErrors(stderr, SDDS_VERBOSE_PrintErrors);
exit(1);
}
if (!SDDS_Terminate(&SDDSout)) {
SDDS_PrintErrors(stderr, SDDS_VERBOSE_PrintErrors);
exit(1);
}
return 0;
}
