刪除在不斷變化的列範圍內具有缺失值的行我有一個數據表，我想在其中刪除可變範圍列的所有列中沒有值的行。
例如，對於下面的數據表：
dt =
| 樣本 | 產品類別 | 第 1 列 | 第2欄 | 第3欄 | 第4欄 | 第5欄 |
| 10001 | A | 0.0122 | 0.0223 | 0.01546 | 0.8742 | 23.12 |
| 10003 | 乙 | | | | | |
| 10004 | A | 0.123 | 0.124 | 0.1240 | 0.9712 | 23.78 |
我想刪除帶有示例“10003”的行，因為它在 Col1 - Col5 中的條目為空。 我知道這個帖子非常相似，但對我來說不是很有幫助，因為我要檢查的空列數通常大於 15，並且列名可以/將會更改，因此手動鍵入此代碼並調整列名
dt << Select Where(
Is Empty(:Col1) &
Is Empty(:Col2) &
is Empty(:Col3) & ... etc.
) << Delete Rows;我可以很容易地得到列表中我需要檢查的所有列是否為空，因為它們都是/將是連續的
cols = dt << Get Column Names(Continuous);
但這就是我陷入困境的地方，我不確定如何繼續。 我是否使用 for 循環來遍歷每一列並檢查它是否為空？ 但我只想刪除列範圍內所有列均為空的行。 如果我使用Python，我只會使用pandas.DataFrame.dropna(axis=0)。 我該如何在 JMP 中執行此操作？
高級統計建模自動化和腳本編寫數據存取大規模定制質量和工藝工程回复：刪除在變化的列範圍內具有缺失值的行所有好的建議都在這裡@tx尼爾森, @ErraticAttack和@gzmorgan0 。
我可以提供我認為最基本的 JSL 命令用法，我認為這些命令足夠強大，可以在不同情況下完成任務。
本質上，我是手動完成的，只做了最小的腳本修改以將其推廣到 JSL 中。
想知道其他人的想法。
dt = Open( "$SAMPLE_DATA/Missing Data Pattern.jmp" );
// this data set actually doesn't have any row that is completely missing.
dt << add rows(3); // now it does
// now the work begins
colNames = dt << get column names( continuous, string );
// make a missing data pattern table
missingdt = dt << Missing Data Pattern(
columns( eval (colNames) ),
Count Missing Value Codes,
Add Value Colors Property,
Output Table( "Missing Data Pattern" )
);
// the next two steps should be fast enough in almost any size of data set that i do not think it is worth writing it as a condition for the cases that no row is completely missing.
// select the row representing all missing. if no such row exists nothing will be selected.
missingdt << select where (:Number of columns missing == N items (colNames));
wait (0);
// delete the selected rows from the original table (if anything was selected)
dt << delete rows;
close (missingdt, no save);
回复：刪除在變化的列範圍內具有缺失值的行你好@cxu04 ,
兩個都@tx尼爾森和@ErraticAttack提供了優秀的解決方案。@ErraticAttack具有選擇排除標準、1 列缺失或全部缺失以及兩者之間的任何值的額外好處。下面，我附上了使用矩陣函數的替代解決方案。僅供參考，一些 JMP 平台分析和圖表具有排除或包含缺失值以及插補的選項。
Names Default To Here( 1 );
// Open Data Table: Blood Pressure.jmp
// → Data Table( "Blood Pressure" )
dt = Open( "$SAMPLE_DATA/Blood Pressure.jmp" );
dt << add rows( 3 );
colNames = dt << get column names( continuous, string );
mymat = dt << get as matrix( colNames );
//this method counts the number of missing values in your column range for each row
//you can set the threshold number for exclusions, misslimit
// any value is missing, missLimit = 0
// all are missing, misslimit = nitems( colNames ) -1
// 20% + misslimit = Ceiling( nitems( colNames)*0.2 ) ) - 1
//
misslimit = 3;
xx = Loc( V Sum( Is Missing( Transpose( mymat ) ) ) > misslimit ) ; //get the row numbers where more than misslimit are missing.
dt << select rows( xx );
dt << exclude();
dt << clear select;
//look at the log output to get an idea what these functions are doing
show( mymat, VSum(IsMissing( Transpose(mymat) )), xx);
以下是 show() 命令的 JMP LOG 輸出的一部分。
show( mymat, VSum(IsMissing( Transpose(mymat) )), xx);
/*:
mymat =
[ 182.595915155309 173.550216520736 180.116643270986 173.797442210887 178.416197148117 181.152265967985 171.278300921568 177.570953487335 170.514896659448,
172.512940709438 180.718560564297 180.917546679043 170.201828509985 178.839136233913 176.084997418848 175.161843431476 184.884491085702 187.603078247176,
180.774421425438 188.500473854652 177.384986937421 188.441549950731 175.346240386252 181.924943930192 183.12980493925 182.794049100295 180.121927541929,
180.620679187184 176.718084823659 182.381646560932 176.337805749776 172.773758285362 183.608276048664 183.04930217403 187.31279033708 182.931206367403,
184.202187190097 180.174755903154 176.277045698247 171.91229023858 174.678173508279 175.578954629626 170.259254710046 189.822176301411 182.694806601573,
171.439595422176 181.462075626649 179.517397097984 183.021068043181 175.683661608275 174.15485890414 184.878981656321 182.387718653956 184.169322994033,
174.589230394192 180.330100756984 180.187858248718 183.498290114926 179.622232533597 172.866924129507 184.16481869131 184.653326274105 180.368840760416,
179.109260746268 172.886222586229 182.904992672503 187.881911714776 192.773275959188 169.793497264777 179.098192996471 184.653672194875 170.403578559484,
177.787439187438 192.050244302843 176.785646502378 184.924564594592 172.499108408691 179.464417655732 186.765084181627 184.384583007231 177.427685029706,
180.687489407623 187.037431530346 181.618651921524 181.868587651411 177.245563534856 176.983573711354 183.864008024749 183.43593738674 184.394933320607,
181.804016028632 185.501158177984 178.209310795898 177.186342937671 176.573252982367 177.20643071091 182.640175499696 190.972901519754 174.986928330607,
178.085353285743 183.877324525276 186.575178728884 174.027593116806 174.72327532089 176.115218469266 179.80577644674 181.46178488602 184.070139882305,
186.313144806757 179.931613365546 180.989181249718 171.059060559811 172.873089144473 181.319863949352 179.038752187802 181.459302956398 189.242086812316,
175.347853139956 180.493132509235 177.2578428812 184.517143208058 171.169474537616 181.880353653926 175.727537977595 184.604441269869 169.540004898628,
181.103289178403 174.944992509985 180.642613381863 179.143677395064 185.445956950321 183.622763290159 183.636564430071 173.249587056935 185.21824057543,
181.288352739913 177.663853371957 178.850703910826 178.211632365764 181.685844613992 178.639264502305 188.483289884327 183.125809166649 173.182049720424,
178.777832744608 174.717037207769 188.986597488081 190.546847185169 178.055520662495 176.991735583422 182.282596238225 187.951725739917 170.940897197153,
176.763152449944 186.677919707096 172.45356859494 179.666182707754 182.726891244295 177.924284755616 183.200306257339 190.139279669601 177.823031925273,
175.768608854708 178.12326354079 182.404679249467 176.614597739946 178.320419876662 177.143465976767 175.856743742269 183.549834391574 183.621024712236,
184.167672392284 171.230704316198 182.335148506991 181.334797218108 186.395858538357 183.421182544757 183.517577768589 171.907000056173 178.774574166082,
. . . . . . . . .,
. . . . . . . . .,
. . . . . . . . .];
V Sum(Is Missing(mymat`)) = [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 9 9 9];
xx = [21 22 23];
回复：刪除在變化的列範圍內具有缺失值的行有一個 JMP 內置程序可以解決這個問題——缺失數據模式。無需重新發明輪子
Names Default To Here( 1 );
dt = Open( "$SAMPLE_DATA/Cities.jmp" );
// column names to search for missing values
names = {"CO", "SO2", "NO", "Lead"};
// you can uncomment the following line to use all 'Continuous' columns
//names = dt << Get Column Names( "Continuous" );
// turn the list into a 'Columns( a, b, c )' expression
columns = Substitute( names, {}, Expr( Columns ) );
// run the missing data pattern
Eval( Eval Expr( table = dt << Missing Data Pattern( Expr( name expr( columns ) ) ) ) );
// select rows where all values are missing
Eval( Eval Expr( table << Select Where( :Number of Columns Missing == N Items( Expr( names ) ) ) ) );
Close( table, No Save );
rows = dt << Get Selected Rows();
// if any rows are selected then they have missing values for all columns given, delete
If( N Rows( rows ) > 0, dt << Delete Rows( rows ) );回复：刪除在變化的列範圍內具有缺失值的行這是處理您的問題的一種方法。其他社區成員可能有更好的方法，但這個方法似乎總是可行。我正在使用名為“血壓”的示例數據表。
下面的腳本的作用是生成選擇行所需的 JSL 語句，如果您對數據表中的所有連續列進行求和，並且總和為缺失值，則選擇該行。所以需要構建的目標JSL是
dt << select where(ismissing(sum(as column(\!"BP 8M\!"),as column(\!"BP 8M\!"),as column(\!"BP 8M\!"),as column(\!"BP 8M\!"),as column(\!"BP 8M\!"),as column(\!"BP 8M\!"),as column(\!"BP 8M\!"),as column(\!"BP 8M\!"),as column(\!"BP 8M\!")))==1);
該腳本在字符串變量中創建此代碼，然後使用
Eval( Parse( ...........);
在示例中，我在數據表中創建了 3 個新行，默認情況下這些行缺少值。然後運行該腳本並找到這 3 行。
Names Default To Here( 1 );
dt =
// Open Data Table: Blood Pressure.jmp
// → Data Table( "Blood Pressure" )
Open( "$SAMPLE_DATA/Blood Pressure.jmp" );
dt << add rows(3);
colNames = dt << get column names( continuous, string );
theExpr = "dt << select where(ismissing(sum(as column(\!"" || colNames[1] || "\!")";
For( i = 2, i <= N Items( colNames ), i++,
theExpr = theExpr || ",as column(\!"" || colNames[1] || "\!")"
);
theExpr = theExpr || "))==1);";
Eval( Parse( theExpr ) )
;
