Dans la BD, les cellules vertes sont les résultats que l’on aimerait obtenir.
dt = Current Data Table();
dur_cols = Filter Each( {col_name}, dt << Get Column Names( "String", Continuous ),
Starts With( col_name, "Date début" )
);
num_colnr = Transform Each( {col_name}, dur_cols, Word( -1, col_name ) );
char_cols = dt << Get Column Names ("string", Character);
wait(1);
N_taches = N Items(dur_cols);
for (i = 1, i <= N_taches, i++,
Eval(
Eval Expr(
col = dt << New Column ( "durée" || num_colnr[i], Numeric, "Continuous", Format(":jou:hh:m",8));
col << Formula(
colidx = Loc(dt[row(), char_cols] > Eval(num_colnr[i]))[1];
If ( N Items(colidx) > 0,
val = dur_cols[colidx] - Eval(dur_cols[i]);
If(IsMissing(val),
0
,
val
);
,
0
);
);
)));Merci pour votre aide !A bit lazy solution, but it might work
Names Default To Here(1);
dt = Open("$DOWNLOADS/AAAA.jmp");
// would get only numeric columns, but the provided example table is broken...
all_cols = dt << Get Column Names(/*Numeric, */"String");
// so first we have to fix the datatable
Remove From(all_cols, 1, 2);
For Each({col_name}, all_cols,
Column(dt, col_name) << Set Data Type("Numeric");
Column(dt, col_name) << Set Modeling Type("Continuous");
);
// create list of duration columns
// same amount of columns AND in same order for "number" and duration
date_cols = {};
some_cols = {};
For Each({col_name}, all_cols,
If(Starts With(lowercase(col_name), "date d"),
Insert Into(date_cols, col_name);
, !IsMissing(Regex(col_name, "^\d+$")),
Insert Into(some_cols, col_name);
,
show(col_name);
);
);
If(N Items(date_cols) != N Items(some_cols),
Throw("Column mismatch");
);
// create new columns
dur_cols = {};
For Each({some_col}, some_cols,
new_col = dt << New Column("duration " || some_col, Numeric, Continuous);
Insert Into(dur_cols, new_col << get name);
);
// calculation
For Each Row(
For Each({date_col, idx}, date_cols,
datesomething = dt[Row(), some_cols[idx]];
datesomethingvals = dt[Row(), some_cols];
// to make comparison easier
datesomethingvals[idx] = .;
datesomethingvals[Loc(datesomethingvals < datesomething)] = .; // lazy handling
idx_of_interest = Loc Min(Abs(datesomethingvals - datesomething));
If(idx_of_interest == 0,
continue();
);
val_of_interest = dt[Row(), date_cols[idx_of_interest]];
dt[Row(), dur_cols[idx]] = Abs(val_of_interest - dt[Row(), date_col]); // not sure if Abs is correct here
);
);
// durations for 900 and 4500 for some reason missing from demo data
Maybe something like this:
For(k=1, k<=6, k++,
StartValue = Column(k+2)[1];
For(i=1, i<=6, i++,
If(num(Column(i+8)[1])>StartValue,break())
);
If(i<7, Write(Column("Date début "||(Column(i+8)<<get name()))[1]-StartValue));
);
Merci pour votre réponse @pauldeen mais il s'agit ici d'une base de données test, ce script devrait fonctionner pour plus de 700 colonnes.
A bit lazy solution, but it might work
Names Default To Here(1);
dt = Open("$DOWNLOADS/AAAA.jmp");
// would get only numeric columns, but the provided example table is broken...
all_cols = dt << Get Column Names(/*Numeric, */"String");
// so first we have to fix the datatable
Remove From(all_cols, 1, 2);
For Each({col_name}, all_cols,
Column(dt, col_name) << Set Data Type("Numeric");
Column(dt, col_name) << Set Modeling Type("Continuous");
);
// create list of duration columns
// same amount of columns AND in same order for "number" and duration
date_cols = {};
some_cols = {};
For Each({col_name}, all_cols,
If(Starts With(lowercase(col_name), "date d"),
Insert Into(date_cols, col_name);
, !IsMissing(Regex(col_name, "^\d+$")),
Insert Into(some_cols, col_name);
,
show(col_name);
);
);
If(N Items(date_cols) != N Items(some_cols),
Throw("Column mismatch");
);
// create new columns
dur_cols = {};
For Each({some_col}, some_cols,
new_col = dt << New Column("duration " || some_col, Numeric, Continuous);
Insert Into(dur_cols, new_col << get name);
);
// calculation
For Each Row(
For Each({date_col, idx}, date_cols,
datesomething = dt[Row(), some_cols[idx]];
datesomethingvals = dt[Row(), some_cols];
// to make comparison easier
datesomethingvals[idx] = .;
datesomethingvals[Loc(datesomethingvals < datesomething)] = .; // lazy handling
idx_of_interest = Loc Min(Abs(datesomethingvals - datesomething));
If(idx_of_interest == 0,
continue();
);
val_of_interest = dt[Row(), date_cols[idx_of_interest]];
dt[Row(), dur_cols[idx]] = Abs(val_of_interest - dt[Row(), date_col]); // not sure if Abs is correct here
);
);
// durations for 900 and 4500 for some reason missing from demo data
Merci pour votre réponse @jthi !
So? Just up the For loop count number. Or make it dynamic by including N cols() as the upper bound.
If you have allready solved this would you please mark this topic as solved?