diff --git a/Scripts/6_recallClassification_Bayes_New_Data.py b/Scripts/6_recallClassification_Bayes_New_Data.py
index 4203b149a3f8d88278460d1af474e722e86ebb14..9f99e02aafdc79f59e807ded98bf95d64fe2e469 100644
--- a/Scripts/6_recallClassification_Bayes_New_Data.py
+++ b/Scripts/6_recallClassification_Bayes_New_Data.py
@@ -41,8 +41,16 @@ def selectFeatures(train_set, train_text, k):
#words = [wnl.lemmatize(t, 'v') for t in words]
print str(len(words))+' words found.'
+ clean_words = []
+ for word in words:
+ try:
+ v = nltk.pos_tag(word)
+ clean_words.append(word)
+ except:
+ continue
+
# Tagging of the parts of speech
- tagged_words = nltk.pos_tag(words)
+ tagged_words = nltk.pos_tag(clean_words)
words = [word for (word, tag) in tagged_words if tag in ['NN','JJ','NNS','RB','VB','VBD','VBG','VBN','VBP','VBZ']]
#words = set(words).intersection(set(unigrams+ bigrams+trigrams));
# Get the frequency of words
diff --git a/Scripts/6_recallClassification_Bayes_New_Data.pyc b/Scripts/6_recallClassification_Bayes_New_Data.pyc
index 5a35e40d8ad44f02be27b0179d58da2e1f431bfb..75ba02db0ddfebcbee448e18e9fb9cc89631374e 100644
Binary files a/Scripts/6_recallClassification_Bayes_New_Data.pyc and b/Scripts/6_recallClassification_Bayes_New_Data.pyc differ
diff --git a/Scripts/7_automatedSteps.py b/Scripts/7_automatedSteps.py
index 49b738be890a9a5b7756040b9c738e47c9a34929..e9411e30c9c11cddff923fb89d9bae9af2d7a7ba 100644
--- a/Scripts/7_automatedSteps.py
+++ b/Scripts/7_automatedSteps.py
@@ -10,7 +10,7 @@ import os
#markers for which sections to execute
pieces = {"Retrieve": False, "Unique": True, "ClassifyS1": True,
- "ClassifyS2": False, "Procodes": True}
+ "ClassifyS2": True, "Procodes": True}
'''
SCRIPT 1 --> Retrieve the data
@@ -23,7 +23,7 @@ if(pieces["Retrieve"]):
basepath = './../Original_Data';
os.chdir(basepath)
- #get data from 2005-2012
+ #get data from 2006-2012
for Year in range(2007, 2008):
print 'Year '+str(Year);
startYear = Year;
diff --git a/Unique_Data/best_keywords.txt b/Unique_Data/best_keywords.txt
index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..846f44ce3cf444d08499c7deb6591c210005d70a 100644
--- a/Unique_Data/best_keywords.txt
+++ b/Unique_Data/best_keywords.txt
@@ -0,0 +1,858 @@
+limited,0.00436636013761
+caused,0.0191131741793
+lack,0.0144507671053
+activated,0.0191131741793
+results,0.00154029959698
+alerting,0.0191131741793
+seals,0.00355533068551
+leads,0.00355533068551
+surgical,0.00714835563023
+sleep,0.0191131741793
+hydraulic,0.00355533068551
+skin,0.00355533068551
+ventilator,0.00436636013761
+enteric,0.00355533068551
+integrity,0.00714835563023
+assembled,0.00714835563023
+webster,0.00355533068551
+improperly,0.00355533068551
+compact,0.00355533068551
+larc,0.00355533068551
+batches,0.00355533068551
+involve,0.00355533068551
+kwiktrak,0.00355533068551
+send,0.0191131741793
+subcomponents,0.00355533068551
+environment,0.00355533068551
+program,0.0386307781791
+present,0.00714835563023
+over-penetration,0.00355533068551
+updated,0.00355533068551
+assay,0.0107798878457
+risk,0.0107798878457
+acoustic,0.0191131741793
+frequencies,0.0191131741793
+compacts,0.00355533068551
+probability,0.00355533068551
+graphic,0.00355533068551
+servicing,0.00355533068551
+fall,0.0107798878457
+affect,0.0144507671053
+surgeon,0.00355533068551
+securely,0.00355533068551
+cool,0.00355533068551
+switches,0.0191131741793
+uni-clip,0.00355533068551
+level,2.76963555943e-05
+tear,0.0107798878457
+issue,0.00436636013761
+shock,0.00436636013761
+list,0.00714835563023
+applicable,0.0191131741793
+standards,0.00436636013761
+leave,0.00355533068551
+assignments,0.0191131741793
+salute,0.00355533068551
+wearer,0.00355533068551
+section,0.00714835563023
+small,0.0107798878457
+integra,0.00355533068551
+prevent,0.0386307781791
+unexpected,0.00355533068551
+antibiotic,0.00355533068551
+portable,0.00355533068551
+bock,0.00355533068551
+preface,0.00355533068551
+direct,0.0045444566034
+pulse,0.0191131741793
+rate,0.0166978648319
+design,0.0168999097695
+insertion,0.00714835563023
+pass,0.00355533068551
+patient,0.000566061730296
+sterile,0.00714835563023
+calibrators,0.00355533068551
+even,0.0492165535006
+established,0.00355533068551
+errors,0.0191131741793
+thyroglobulin,0.00355533068551
+saved,0.00355533068551
+specification,0.0219140668989
+selected,0.0386307781791
+remain,0.00436636013761
+thickness,0.00355533068551
+access,0.008929713772
+version,0.0585735750948
+international,0.00355533068551
+mammography,0.0191131741793
+unintentionally,0.0191131741793
+ignition,0.00355533068551
+non-linear,0.00355533068551
+construct,0.00355533068551
+taut,0.00355533068551
+degree,0.0191131741793
+exchange,0.00355533068551
+component,0.0144507671053
+indicator,0.00355533068551
+ventilators,0.0191131741793
+reported,0.0168999097695
+pause,0.0191131741793
+threads,0.00355533068551
+latex-free,0.00355533068551
+directions,2.76963555943e-05
+based,0.00355533068551
+perforation,0.00355533068551
+monitor-defibrillator,0.0191131741793
+change,0.008929713772
+extreme,0.00355533068551
+engage,0.00355533068551
+taper,0.00714835563023
+technical,0.0191131741793
+shift,0.00714835563023
+terminals,0.00355533068551
+reports,0.00714835563023
+prior,0.00355533068551
+resulting,2.76963555943e-05
+smoke,0.00355533068551
+inaccurage,0.00355533068551
+anaerotest,0.00355533068551
+changes,0.00355533068551
+diameter,0.0107798878457
+technique,0.00355533068551
+regardless,0.0191131741793
+valproic,0.00355533068551
+extra,0.00355533068551
+requiring,0.00355533068551
+cylinder,0.00355533068551
+nicoletone,0.0191131741793
+anomalies,0.00355533068551
+atrium,0.00355533068551
+apply,0.00355533068551
+counts,0.00355533068551
+staple,0.00355533068551
+unit,0.00714835563023
+fracturing,0.00355533068551
+overheat,0.00355533068551
+variation,0.00355533068551
+humidifier,0.00355533068551
+contains,0.00355533068551
+connector,0.00714835563023
+process,0.0107798878457
+sample,0.000975330714483
+biliary,0.00355533068551
+biosense,0.00355533068551
+memory,0.0168999097695
+taken,0.00355533068551
+black,0.00355533068551
+panel,0.00355533068551
+minor,0.00355533068551
+separated,0.00355533068551
+plug,0.00714835563023
+modification,0.00714835563023
+initiated,0.00355533068551
+nebulizers,0.00355533068551
+visual,0.0191131741793
+normalized,0.00355533068551
+central,0.0191131741793
+hole,2.76963555943e-05
+pouch,0.00355533068551
+clumps,0.00355533068551
+customer,0.00714835563023
+high,0.0144507671053
+testying,0.00355533068551
+anaerobic,0.00355533068551
+recall,0.00355533068551
+positives-the,0.00355533068551
+modified,0.00355533068551
+labeled,0.00263623971185
+values,0.00355533068551
+temperatures,0.00355533068551
+drill,0.00355533068551
+following,0.00436636013761
+meet,0.0107798878457
+male,0.00355533068551
+control,0.0322635401109
+bent,0.00355533068551
+hemoglobin,0.00355533068551
+lock,0.00714835563023
+chip,0.00436636013761
+bennett,0.00355533068551
+catheters,0.0144507671053
+states,0.00714835563023
+indicates,0.00714835563023
+detaching,0.00714835563023
+defibrillation,0.00436636013761
+numbers,0.00355533068551
+serial,0.00355533068551
+occur,0.0585735750948
+counterfeit,0.0107798878457
+information,0.00355533068551
+triathlon,0.00355533068551
+recovery,0.00436636013761
+defective,0.00436636013761
+provide,0.00714835563023
+iggt,0.00355533068551
+returning,0.0191131741793
+feature,0.0386307781791
+delay,0.0386307781791
+gate,0.00355533068551
+parameter,0.0386307781791
+profile,0.00355533068551
+product,0.0139063901664
+specifications,0.0144507671053
+stop,0.0208557213037
+containing,0.00355533068551
+produce,0.0191131741793
+fatigue,0.00355533068551
+designed,0.0386307781791
+date,0.00714835563023
+hemostatic,0.00355533068551
+data,0.0191131741793
+response,0.00436636013761
+workstation,0.0191131741793
+overspeed,0.0191131741793
+laminate,0.00355533068551
+microcoagulation,0.0191131741793
+scenarios,0.0191131741793
+switch,0.00436636013761
+used,0.0137046985356
+allow,2.76963555943e-05
+correlation,0.00355533068551
+infant,0.00355533068551
+deform,0.00355533068551
+burn,0.00355533068551
+enter,0.00714835563023
+wound,0.00355533068551
+move,0.000975330714483
+displayed,0.0386307781791
+unacceptable,0.0191131741793
+damage,0.000975330714483
+stability,0.00355533068551
+thermal,0.00355533068551
+discrepancy,0.0386307781791
+solder,0.00355533068551
+contaminated,0.00355533068551
+regions,0.0191131741793
+outer,0.00714835563023
+inlet,0.00714835563023
+susceptible,0.00436636013761
+lateral,0.00355533068551
+actually,0.0181618611395
+forms,0.00355533068551
+detected,0.00355533068551
+production,0.00355533068551
+condition,0.0492165535006
+communication,0.0191131741793
+sterility,0.0181618611395
+renaming,0.0191131741793
+affected,0.00714835563023
+sterilization,0.00355533068551
+auto,0.0386307781791
+handpiece,0.00355533068551
+material,0.0181618611395
+processes,0.00355533068551
+break,0.0181618611395
+reagent,0.00714835563023
+cutting,0.00355533068551
+therapy,0.0191131741793
+undersized,0.00714835563023
+foot,0.00355533068551
+rotation,0.00436636013761
+detach,0.0107798878457
+silver,0.00355533068551
+dielectric,0.00355533068551
+instructions,0.00714835563023
+stopper,0.00355533068551
+chamfer,0.00355533068551
+oxygen,0.00355533068551
+chronic,0.00355533068551
+edge,0.00355533068551
+drop,0.00355533068551
+identified,0.0168999097695
+labeling,0.0107798878457
+sufficiently,0.00355533068551
+bolts,0.00355533068551
+vibrations,0.00355533068551
+found,0.0107798878457
+tray,0.0107798878457
+erratic,0.00355533068551
+bone,0.00714835563023
+upward,0.00355533068551
+baxter,0.0191131741793
+interruption,0.0191131741793
+servo-i,0.00436636013761
+activate,0.0191131741793
+accelerator,0.0191131741793
+fistula,0.00355533068551
+energy,0.0386307781791
+receipt,0.00355533068551
+related,0.00355533068551
+wires,0.00355533068551
+connect,0.00263623971185
+calibrate,0.00355533068551
+resuscitate,0.0191131741793
+separating,0.00355533068551
+laboratory,0.0191131741793
+operation,0.0322635401109
+event,0.0585735750948
+standard,0.00436636013761
+large,0.0191131741793
+cubicin,0.00355533068551
+display,0.143083675196
+electronic,0.0191131741793
+valve,0.00355533068551
+interference,0.0191131741793
+imprecision,0.00714835563023
+contained,0.00355533068551
+investigation,0.00355533068551
+safety,0.00714835563023
+print,0.0191131741793
+cause,0.0135509338379
+occurs,0.0585735750948
+assurance,0.0107798878457
+shows,0.00355533068551
+linear,0.000975330714483
+barrier,0.00355533068551
+workflow,0.0191131741793
+centricity,0.0191131741793
+possibility,0.00355533068551
+promptly,0.00355533068551
+difficulty,0.00355533068551
+reason,0.0191131741793
+base,0.00714835563023
+bayer,0.00355533068551
+earliest,0.00355533068551
+generate,0.00436636013761
+wall,0.00355533068551
+care,0.00355533068551
+incorporates,0.00355533068551
+exposing,0.00355533068551
+programming,0.0191131741793
+modules,0.0191131741793
+filter,0.0107798878457
+turn,0.0168999097695
+length,0.00714835563023
+-revised,0.00355533068551
+unknown,0.0107798878457
+conducted,0.00355533068551
+cleaning,0.00355533068551
+carryover,0.00355533068551
+distal,0.00355533068551
+number,0.0181618611395
+trocars,0.00355533068551
+reacher,0.00355533068551
+channels,0.0191131741793
+instances,0.00355533068551
+approximately,0.00714835563023
+message,0.0673140524201
+quality,0.0107798878457
+size,0.0219140668989
+given,0.00355533068551
+temperature,0.0168999097695
+service,0.0168999097695
+become,0.000975330714483
+leading,0.00355533068551
+system,0.0352750783848
+elevator,0.00355533068551
+stimulation,0.00355533068551
+compromise,0.00714835563023
+platlet,0.00355533068551
+lumen,0.00355533068551
+fully-extended,0.0191131741793
+recent,0.00714835563023
+bits,0.00355533068551
+ranging,0.00355533068551
+recovered,0.00355533068551
+glucose,0.00355533068551
+ceiling,0.00355533068551
+thaw,0.00355533068551
+amplifiers,0.0191131741793
+bowel,0.00355533068551
+needle,0.00355533068551
+prosthetic,0.00714835563023
+i-stat,0.00355533068551
+compressor,0.00355533068551
+prosthesis,0.00355533068551
+part,0.0181618611395
+cartridges,0.00355533068551
+v-lead,0.00355533068551
+detection,0.0191131741793
+cannula,0.00714835563023
+target,0.00355533068551
+bubble,0.0191131741793
+preparation,0.00355533068551
+patients,0.0107798878457
+determined,0.00714835563023
+architect,0.00436636013761
+transducer,0.0191131741793
+assays,0.00355533068551
+accessed,0.0191131741793
+treated,0.00355533068551
+exhibit,0.0107798878457
+false,0.0045444566034
+de-lamination,0.00355533068551
+correction,0.00436636013761
+intraocular,0.00355533068551
+turned,0.0191131741793
+defect,0.000975330714483
+left-handed,0.00355533068551
+medical,0.00355533068551
+screws,0.00355533068551
+concentration,0.00355533068551
+neeg,0.0191131741793
+sell,0.00355533068551
+strength,0.00714835563023
+built,0.00355533068551
+analyzers,0.00714835563023
+able,0.0065905709788
+mechanism,0.00355533068551
+contact,0.0168999097695
+also,0.00355533068551
+chest,0.00355533068551
+potential,0.00311589180703
+flags,0.0191131741793
+knobs,0.0191131741793
+milled,0.00355533068551
+completed,0.00355533068551
+channel,0.0191131741793
+blades,0.00355533068551
+track,0.00355533068551
+denominator,0.0191131741793
+correct,0.000568903190553
+systems,0.0386307781791
+alarms,0.0191131741793
+tilt,0.00355533068551
+lists,0.00355533068551
+significant,0.00355533068551
+device,0.0453449123168
+knee,0.00355533068551
+embolization,0.00355533068551
+mach,0.00355533068551
+latex,0.00355533068551
+polyethylene,0.00355533068551
+prolongation,0.00355533068551
+defect-the,0.00355533068551
+average,0.00355533068551
+lose,0.0191131741793
+metal,0.00714835563023
+drive,0.00714835563023
+dissolved,0.00355533068551
+uroview,0.00355533068551
+clearing,0.0191131741793
+professional,0.0191131741793
+left,0.00714835563023
+clinician,0.0191131741793
+fact,0.0295455555902
+slope,0.00355533068551
+selection,0.0191131741793
+labels,0.0191131741793
+friction,0.00355533068551
+rotating,0.00355533068551
+elevate,0.000260952805506
+microalbumin,0.00355533068551
+staff,0.00355533068551
+electromagnetic,0.0386307781791
+upgrade,0.0191131741793
+ratio,0.0238963205875
+distributed,0.00355533068551
+nebulization,0.00355533068551
+failed,0.00355533068551
+tears,0.00714835563023
+unable,0.00355533068551
+mislabeled-,0.00355533068551
+employed,0.00355533068551
+partially,0.00355533068551
+plate,0.0107798878457
+tina-quant,0.00355533068551
+radiated,0.0191131741793
+dependent,0.00355533068551
+overpressure,0.0191131741793
+biopsy,0.0107798878457
+joint,0.00714835563023
+october,0.0191131741793
+breast,0.00714835563023
+report,0.0137046985356
+chips,0.00355533068551
+comply,0.0191131741793
+disabled,0.0191131741793
+regarding,0.00714835563023
+symbia,0.0191131741793
+sleeve,0.00355533068551
+settings,0.0191131741793
+sirolimus,0.00355533068551
+experience,0.0191131741793
+gloves,0.00355533068551
+contain,0.0144507671053
+release,0.0191131741793
+movement,0.00714835563023
+gantry,0.0191131741793
+exists,0.00436636013761
+acquired,0.0191131741793
+testing,0.0107798878457
+module,0.000975330714483
+freeze,0.00355533068551
+observed,0.00714835563023
+dislodge,0.00355533068551
+intensity,0.0191131741793
+individual,0.00355533068551
+result,0.00783938249494
+seal,0.0181618611395
+fail,0.00784669525049
+pinhole,0.0191131741793
+surgery,0.00355533068551
+electrode,0.00355533068551
+heaters,0.00355533068551
+fails,0.00436636013761
+capacity,0.00714835563023
+unimplanted,0.00355533068551
+attempting,0.00355533068551
+lots,0.0144507671053
+subsequent,0.00355533068551
+currently,0.00355533068551
+label,0.0056123041309
+pump,0.0678517154754
+exhalation,0.00355533068551
+correctly,0.00120628124022
+c-arm,0.00355533068551
+affecting,0.00355533068551
+corruption,0.0191131741793
+oversized,0.00355533068551
+emissions,0.0191131741793
+available,0.00355533068551
+recently,0.00355533068551
+opening,0.00355533068551
+packaged,0.00355533068551
+flow,0.00436636013761
+analyzer,0.000975330714483
+however,0.00355533068551
+comm,0.0386307781791
+interface,0.0386307781791
+disconnect,0.00714835563023
+packages,0.00714835563023
+come,0.000635028600811
+roller,0.0322635401109
+monitor,0.0322635401109
+needs,0.0386307781791
+received,0.0181618611395
+installation,0.0191131741793
+swan-ganz,0.00355533068551
+limits,0.0386307781791
+fault,0.0386307781791
+tapered,0.00355533068551
+drug,0.00355533068551
+foreign,0.00355533068551
+connection,0.00714835563023
+veeg,0.0191131741793
+operating,0.0191131741793
+lenses,0.00355533068551
+adjusted,0.00355533068551
+compromising,0.00714835563023
+tone,0.0386307781791
+pose,0.0144507671053
+battery,0.00436636013761
+invasive,0.0191131741793
+excess,0.00436636013761
+height,0.00355533068551
+packaging,0.0144507671053
+table,0.00201194745386
+strong,0.00355533068551
+cycling,0.0191131741793
+raise,0.00355533068551
+re-designed,0.0191131741793
+mounting,0.00355533068551
+layer,0.00355533068551
+quickly,0.0191131741793
+reduction,0.00355533068551
+slowly,0.00355533068551
+certain,0.000260952805506
+expected,2.76963555943e-05
+protrusions,0.00355533068551
+utilizing,0.0191131741793
+anomaly,0.0191131741793
+board,0.0191131741793
+firm,0.00201194745386
+direction,0.0045444566034
+versa,0.00355533068551
+efficacy,0.00355533068551
+packs,0.00436636013761
+fire,0.00355533068551
+concerning,0.00355533068551
+precision,0.0107798878457
+readings,0.00355533068551
+controller,0.0191131741793
+lift,0.0144507671053
+external,0.00714835563023
+protocols,0.0191131741793
+sound,0.0191131741793
+shaft,0.00355533068551
+multi,0.00355533068551
+disconnected,0.00355533068551
+pacs,0.0191131741793
+frozen,0.0191131741793
+conformance,0.0191131741793
+mount,0.00355533068551
+resume,0.0191131741793
+corrode,0.00355533068551
+employ,0.00714835563023
+abrupt,0.00355533068551
+expiration,0.00355533068551
+inaccurately,0.00355533068551
+error,0.018707551114
+stopping,0.0191131741793
+situation,0.00436636013761
+transfering,0.0191131741793
+guide,0.00355533068551
+procedure,0.00355533068551
+welding,0.00355533068551
+collimator,0.0191131741793
+diaphragm,0.00355533068551
+injury,0.00714835563023
+coating,0.00355533068551
+tests,0.0191131741793
+surface,0.00355533068551
+bias,0.00355533068551
+turbidimetric,0.00355533068551
+malformed,0.00355533068551
+joints,0.00714835563023
+confirmed,0.0386307781791
+permeability,0.00355533068551
+compromised,0.00355533068551
+breakage-memory,0.00355533068551
+radiation,0.0191131741793
+material-,0.00355533068551
+elevated,2.76963555943e-05
+prothrombin,0.00355533068551
+shorter,0.00714835563023
+potentially,0.0144507671053
+otto,0.00355533068551
+incorrectly,0.000260952805506
+units,0.00355533068551
+calculation,0.0191131741793
+strips,0.00714835563023
+inadvertently,0.0386307781791
+numerical,0.0191131741793
+driver,0.00355533068551
+effect,0.00714835563023
+alert,0.0585735750948
+action,0.00355533068551
+running,0.00436636013761
+levels,0.00355533068551
+uses,0.0191131741793
+user,0.0350429690939
+catheter,0.0181618611395
+cycle,0.00436636013761
+threaded,0.00355533068551
+suppliers,0.00355533068551
+studies,0.00355533068551
+calibration,0.0386307781791
+analysis,0.00355533068551
+alsius,0.00355533068551
+materials,0.00355533068551
+acid,0.00355533068551
+components,0.00714835563023
+inability,0.00714835563023
+position,0.0191131741793
+model,0.00355533068551
+laboratories,0.00355533068551
+protege,0.00355533068551
+bending,0.00355533068551
+assembling,0.00355533068551
+protocol,0.0191131741793
+storage,0.00714835563023
+rest,0.00355533068551
+fibrin,0.00355533068551
+identify,0.00714835563023
+locking-screws,0.00355533068551
+proximal,0.00355533068551
+speed,0.0386307781791
+cw-mode,0.0191131741793
+console,0.00355533068551
+cobas,0.00355533068551
+generator,0.0191131741793
+stent,0.00355533068551
+exceed,0.0191131741793
+pallet,0.0191131741793
+instrument,0.000975330714483
+causing,0.00201194745386
+treatment,0.00436636013761
+samples,0.00714835563023
+non-operational,0.0191131741793
+apart,0.00714835563023
+remaining,0.0191131741793
+match,0.00355533068551
+complaint,0.00120628124022
+elevation,0.00355533068551
+increased,0.00436636013761
+loss,0.00355533068551
+specimen,0.00714835563023
+early,0.0191131741793
+misbranding-one,0.00355533068551
+performance,0.0191131741793
+using,0.00311589180703
+breakage,0.00714835563023
+essential,0.0191131741793
+insert,0.0144507671053
+modifications,0.00355533068551
+instead,0.00355533068551
+ring,0.00233900939652
+manual,0.00436636013761
+resin,0.00355533068551
+manifest,0.0191131741793
+specific,0.0257083118903
+use-puritan,0.00355533068551
+continue,0.0191131741793
+unstable,0.00355533068551
+fully,0.0386307781791
+output,0.0386307781791
+slings,0.00355533068551
+reduced,0.0107798878457
+therapeutic,0.00355533068551
+thermodilution,0.00355533068551
+seating,0.00355533068551
+diluent,0.0191131741793
+motor,0.0191131741793
+exposed,0.0107798878457
+cosycot,0.00355533068551
+moving,0.00355533068551
+back,0.00714835563023
+warmer,0.00355533068551
+specified,0.0191131741793
+images,0.0191131741793
+intact,0.00355533068551
+lead,0.0334267909414
+notification,0.0191131741793
+tube,0.00714835563023
+carabineer,0.00355533068551
+results-protime,0.0191131741793
+cradle,0.0191131741793
+provides,0.00355533068551
+demonstrated,0.00355533068551
+assembly,0.00355533068551
+inaccurate,0.0107798878457
+power,0.000975330714483
+manufactured,0.0107798878457
+processing,0.00355533068551
+pressure,0.0585735750948
+step,0.00355533068551
+shavings,0.00355533068551
+bolt,0.00714835563023
+tubing,0.0107798878457
+losing,0.0191131741793
+post,0.00355533068551
+manufacturing,0.00355533068551
+manufacturer,0.0107798878457
+heparin,0.00355533068551
+length-cannulated,0.00355533068551
+actual,0.0044230932165
+package,0.0334267909414
+column,0.00355533068551
+soft,0.0998274758132
+numerator,0.0191131741793
+introducer,0.00355533068551
+audible,0.0386307781791
+range,0.0191131741793
+dialog,0.0191131741793
+mixed,0.00355533068551
+introduced,0.0191131741793
+antibody,0.00355533068551
+automatically,0.0191131741793
+software,0.0789642112942
+malfunction,0.00355533068551
+instruments,0.00355533068551
+image,0.0191131741793
+functional,0.0191131741793
+complaints,0.0107798878457
+false-positive,0.00355533068551
+operator,0.0191131741793
+pivot,0.00355533068551
+cardioplegia,0.00355533068551
+cleared,0.0386307781791
+computer,0.0191131741793
+support,0.00355533068551
+calibrator,0.000975330714483
+specifically,0.00355533068551
+custom,0.0107798878457
+width,0.00355533068551
+clinical,0.0191131741793
+inner,0.0144507671053
+sealed,0.00355533068551
+series,0.0191131741793
+albumin,0.00714835563023
+specimens,0.00714835563023
+function,0.0208557213037
+becoming,0.00355533068551
+delivery,0.0191131741793
+failure,0.0135509338379
+heat,0.0181618611395
+properly,0.0107798878457
+recoil,0.00355533068551
+technician,0.0191131741793
+radiopaque,0.00355533068551
+reset,0.0386307781791
+made,0.00355533068551
+versions,0.0191131741793
+default,0.0191131741793
+inside,0.00355533068551
+attached,0.00355533068551
+maximum,0.0386307781791
+devices,0.0144507671053
+different,0.00714835563023
+placed,0.00355533068551
+customers,0.00355533068551
+reorder,0.0191131741793
+strands,0.00355533068551
+alarm,0.0998274758132
+bearing,0.00355533068551
+required,0.00355533068551
+associated,0.00436636013761
+dual,2.76963555943e-05
+defined,0.00355533068551
+inadequate,0.000975330714483
+mislabeled,0.00714835563023
+metals,0.00355533068551
+resultant,0.00355533068551
+proper,0.0144507671053
+exist,0.000975330714483
+request,0.00355533068551
+sheath,0.00714835563023
+tibial,0.00355533068551
+internal,0.00436636013761
+counterfeit-blood,0.00355533068551
+incorrect,0.000861756191233
+compared,0.00436636013761
+floor,0.00355533068551
+alerts,0.0191131741793
+generally,0.00355533068551
+enclosure,0.00355533068551
+field,0.00355533068551
+falsely,0.00436636013761
+electrical,0.00355533068551
+test,0.0044230932165
+antimicrobial,0.00355533068551
+implant,0.0107798878457
+monitors,0.00355533068551
+requested,0.00355533068551
+repeat,0.0191131741793
+users,0.00355533068551
+separate,0.0144507671053
+drainage,0.00355533068551
+problems,0.0191131741793
+senographe,0.0191131741793
+luer,0.00355533068551
+allowing,0.00714835563023
+rotatinghemostatic,0.00355533068551
+aware,0.00355533068551
+exposure,0.0191131741793
+longer,0.00355533068551
+vice,0.0056123041309
+interval,0.00355533068551
+holes,0.00714835563023
+immunoglobulin,0.00355533068551
+incomplete,0.00714835563023
+time,0.0168999097695
+failures,0.000975330714483
+validation,0.00714835563023
+starting,0.0191131741793
+decision,0.00355533068551
+rinse,0.00355533068551
+valves,0.00355533068551
diff --git a/Unique_Data/unique2007_classified.xls b/Unique_Data/unique2007_classified.xls
new file mode 100644
index 0000000000000000000000000000000000000000..14eeb33014b7557d33a7998f9e365cff4781fd87
Binary files /dev/null and b/Unique_Data/unique2007_classified.xls differ