fix typos in comments (#7)

Author: avec-zz

em.q

@@ -15,7 +15,7 @@ mf:.ml.wbinmle[n;0]          / parameter maximization function
 phi:2#1f%2f                  / coins are picked with equal probability
 .ml.em[1b;lf;mf;x] . pT:(phi;flip enlist THETA)
 (.ml.em[1b;lf;mf;x]//) pT  / call until convergence
-/ which flips came from which THETA? pick maximum log likelkhood
+/ which flips came from which THETA? pick maximum log likelihood
 
 pT:(.ml.em[1b;lf;mf;x]//) pT
 .util.assert[1 0 0 1 0] .ml.imax .ml.likelihood[0b;lf;x] . pT
@@ -46,7 +46,7 @@ mu0:10 20 30                    / distribution's mu
 s20:s0*s0:1 3 2                 / distribution's variance
 m0:100 200 150                  / number of points per distribution
 X:raze X0:mu0+s0*(.ml.bm ?[;1f]::) each m0 / build dataset
-show .util.plt raze each (X0;0f*X0),'(X0;.ml.gaussl'[mu0;s20;X0]) / plot 1d data and guassian curves
+show .util.plt raze each (X0;0f*X0),'(X0;.ml.gaussl'[mu0;s20;X0]) / plot 1d data and gaussian curves
 k:count mu0
 phi:k#1f%k;      / guess that distributions occur with equal frequency
 mu:neg[k]?X;     / pick k random points as centers

fmincg.q

@@ -1,4 +1,4 @@
-/ Minimize a continuous differentialble multivariate function. Starting point
+/ Minimize a continuous differentiable multivariate function. Starting point
 / is given by "X" (D by 1), and the function named in the string "f", must
 / return a function value and a vector of partial derivatives. The Polack-
 / Ribiere flavour of conjugate gradients is used to compute search
@@ -53,7 +53,7 @@
 /  f) introduced BREAK variable to overcome q's lack of break statement
 / 2) max length "n" is now mandatory
 
-\d .fmincg / function minimize nonlinear congjugate gradient
+\d .fmincg / function minimize nonlinear conjugate gradient
 
 RHO:.01     / a bunch of constants for line searches
 SIG:.5      / RHO and SIG are the constants in the Wolfe-Powell conditions
@@ -93,12 +93,12 @@ minimize:{[F;v]
 extrapolate:{[F;v]
  v[`z2]:cubicextrapolation . v`f2`f3`d2`d3`z3;
  v[`z2]:$[$[0>v`z2;1b;0w=v`z2];$[.5>=v`limit;v[`z1]*EXT-1;.5*v[`limit]-v`z1];
-  / extraplation beyond max? -> bisect
+  / extrapolation beyond max? -> bisect
   $[-.5<v`limit;v[`limit]<v[`z2]+v`z1;0b];.5*v[`limit]-v`z1;
-  / extraplation beyond limit? -> set to limit
+  / extrapolation beyond limit? -> set to limit
   $[-.5>v`limit;(EXT*v`z1)<v[`z2]+v`z1;0b];v[`z1]*EXT-1;
   v[`z2]<v[`z3]*neg INT;v[`z3]*neg INT;
-  / too clost to limit?
+  / too close to limit?
   $[-.5<v`limit;v[`z2]<(v[`limit]-v`z1)*1f-INT;0b];(v[`limit]-v`z1)*1f-INT;
   v[`z2]];
  v[`f3]:v`f2;v[`d3]:v`d2;v[`z3]:neg v`z2; / set pt 3 = pt 2

ml.q

@@ -400,7 +400,7 @@ gaussl:{[mu;sigma;x]
  p:exp (x*x-:mu)%-2*sigma;
  p%:sqrt sigma*twopi;
  p}
-/ guassian log likelihood
+/ gaussian log likelihood
 gaussll:{[mu;sigma;X] -.5*sum (logtwopi;log sigma;(X*X-:mu)%sigma)}
 / gaussian maximum likelihood estimator
 gaussmle:{[x](mu;avg x*x-:mu:avg x)}
@@ -433,7 +433,7 @@ likelihood:{[l;lf;X;phi;THETA]
  p:$[l;p+log phi;p*phi];       / apply prior probabilities
  p}
 
-/ using (l)ikelhood (f)unction, (w)eighted (m)aximum likelihood estimator
+/ using (l)ikelihood (f)unction, (w)eighted (m)aximum likelihood estimator
 / (f)unction with prior probabilities (p)hi and distribution parameters
 / (THETA) (with optional (f)itting of (p)hi) perform expectation maximization
 em:{[fp;lf;wmf;X;phi;THETA]
@@ -448,7 +448,7 @@ em:{[fp;lf;wmf;X;phi;THETA]
 tdm:{[c;v](-1_@[(1+count v)#0;;+;1]::) each v?c}
 
 lntf:{log 1f+x}                    / log normalized term frequency
-dntf:{[k;x]k+(1f-k)*x% max each x} / double normalized term frequenecy
+dntf:{[k;x]k+(1f-k)*x% max each x} / double normalized term frequency
 
 idf: {log count[x]%sum 0<x}     / inverse document frequency
 idfs:{log 1f+count[x]%sum 0<x}  / inverse document frequency smooth
@@ -554,7 +554,7 @@ dt:{[cgf;ogf;ipf;opt;w;t]
  opt:(`maxd`minss`minsl`ming`maxff!(0N;2;1;0;::)),opt; / default options
  if[0=opt`maxd;:(w;first d)];    / check if we've reached max depth
  if[identical a:first d;:(w;a)]; / check if all values are equal
- if[opt[`minss]>count a;:(w;a)]; / check if insufficent samples
+ if[opt[`minss]>count a;:(w;a)]; / check if insufficient samples
  d:((neg floor opt[`maxff] count d)?key d)#d:1 _d;   / sub-select features
  d:{.[x isord z;y] z}[(cgf;ogf);(ipf;w;a)] peach d;  / compute gains
  d:(where (any opt[`minsl]>count each last::) each d) _ d; / filter on minsl
@@ -575,7 +575,7 @@ dtcr:{[tr;d]                    / recursive component
  v:(,'/) tr[2] .z.s\: d;    / dig deeper for null values
  v}
 
-/ decistion tree pruning primitives
+/ decision tree pruning primitives
 
 / wilson score - binary confidence interval (Edwin Bidwell Wilson)
 wscore:{[z;f;n](f+(.5*z2n)+-1 1f*z*sqrt((.25*z2n)+f-f*f)%n)%1f+z2n:z*z%n}
@@ -637,7 +637,7 @@ dtkfxv:{[dtf;ef;a;ts]kfxvt[dtmincc[ef]\[;a]dtf::;dtc\:/:;ts]}
 
 / decision tree utilities
 
-/ print leaf: prediction followd by classification error% or regresssion sse
+/ print leaf: prediction followed by classification error% or regression sse
 pleaf:{[w;x]
  v:waom[w;x];                   / value
  e:$[isord x;string sum e*e:v-x;string[.1*"i"$1e3*1f-avg x = v],"%"];
@@ -693,7 +693,7 @@ rt:dt[oig;oig;wmse]             / regression tree
 / random forest
 
 / generate (n) decision trees by applying (f) to a resampled (with
-/ replacemnt) (t)able
+/ replacement) (t)able
 bag:{[n;f;t](f ?[;t]::) peach n#count t} / (b)ootstrap (ag)gregating
 
 / given an atom or list (k), and bootstrap aggregating (m)odel, make

porter.q

@@ -38,15 +38,15 @@ hasvowel:{any vowel x}
 / returns true if x ends in a double consonant
 doublec:{$[2>count x;0b;(=) . -2#x;last cons x;0b]}
 
-/ return true if last three letters are consontant - vowel -
-/ consontant and last letter is not in "wxy"
+/ return true if last three letters are consonant - vowel -
+/ consonant and last letter is not in "wxy"
 cvc:{$[3>count x;0b;101b~-3#cons x;not last[x] in "wxy";0b]}
 
-/ if a<m replace n charaters with (r)eplacement suffix
+/ if a<m replace n characters with (r)eplacement suffix
 r:{[a;n;r;x]$[a<m n:n _ x;n,r;x]}
 
 / compute m where m in c?(vc){m}v? and c and v are consecutive lists
-/ of consontants and vowels
+/ of consonants and vowels
 m:{sum[x] - first x:x where differ x:cons x}
 
 / remove plurals and -ed or -ing